Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Age-related changes in visual encoding strategy preferences during a spatial memory task

Age-related changes in visual encoding strategy preferences during a spatial memory task Ageing is associated with declines in spatial memory, however, the source of these deficits remains unclear. Here we used eye-tracking to investigate age-related differences in spatial encoding strategies and the cognitive processes underlying the age-related deficits in spatial memory tasks. To do so we asked young and older participants to encode the locations of objects in a virtual room shown as a picture on a computer screen. The availability and utility of room-based landmarks were manipulated by removing landmarks, presenting identical landmarks rendering them uninformative, or by presenting unique landmarks that could be used to encode object locations. In the test phase, participants viewed a second picture of the same room taken from the same (0°) or a different perspective (30°) and judged whether the objects occupied the same or different locations in the room. We found that the introduction of a perspective shift and swapping of objects between encoding and testing impaired performance in both age groups. Furthermore, our results revealed that although older adults performed the task as well as younger participants, they relied on different visual encoding strategies to solve the task. Specifically, gaze analysis revealed that older adults showed a greater preference towards a more categorical encoding strategy in which they formed relationships between objects and landmarks. Introduction 2020; Montefinese et al., 2015; Muffato et al., 2019; Segen et al., 2020). Building on these studies, and to gain a more Successful navigation and orientation depend on our abil- detailed understanding of the factors that contribute to the ity to recognise familiar places across different perspectives performance decline, we use eye-tracking to investigate (Waller & Nadel, 2013). In the lab, this ability is typically potential age-related differences in visual encoding strate- assessed with tasks in which participants first encode an gies. Specifically, we are interested in whether young and array of objects or environmental features from one per- older adults rely on the same or different environmental cues spective and are then asked to indicate whether the array during place recognition. has changed when presented from a different perspective. Recently, Muffato et al. (2019) and Hilton et al. (2020) Studies using such paradigms have reported age-related investigated the effects of cognitive ageing on place recog- declines in performance (Hartley et al., 2007; Hilton et al., nition abilities using scenes defined by objects that were placed in an open field. After encoding a scene with four objects, participants were presented with another scene * Vladislava Segen from a different perspective and had to decide whether or vladislava.segen@gmail.com not it was identical to the one encoded. Results revealed the Ageing and Dementia Research Centre, Bournemouth presence of object-location binding errors, particularly in University, Bournemouth, UK older adults. That is, compared to younger participants, older Department of Psychology, Bournemouth University, adults found it harder to detect that two objects had swapped Bournemouth, UK locations than when one of the objects was replaced with a Department of Psychology, University of Cyprus, Nicosia, new object. Cyprus In our previous work (Segen et al., 2020), we investi- CYENS Centre, Nicosia, Cyprus gated age-related differences in the ability to recognise spa- Aging and Cognition Research Group, German Center tial configurations across different perspectives. The task for Neurodegenerative Diseases (DZNE), Magdeburg, required participants to encode the locations of an array Germany Vol:.(1234567890) 1 3 Psychological Research (2022) 86:404–420 405 of identical objects presented as an image on a computer of landmarks. Second, the landmarks used in the environ- screen. The objects were arranged in clusters of one, two ment were all unique and informative and could, therefore, and three objects, in a virtual room containing additional facilitate the encoding of the object locations, even if they environmental cues such as windows and a door. Then, par- distracted the older participants. Third, there was substantial ticipants viewed a second image of the same room taken overlap between the landmarks and the objects of the scene, from the same (0°) or a different perspective (45° or 135°) which prevented a region of interest analysis. Finally, due and judged whether or not the objects were in the same to the large perspective shifts introduced in some trials (e.g. locations. The positions of the objects were either changed 135°), some landmarks were visible during encoding but by swapping two object clusters or by rotating one of the not at test. clusters. While with the former manipulation the task could The current study was designed to disentangle the expla- be solved using a coarse categorical representation of the nations for age-related differences in place recognition by spatial relationships between object clusters (e.g. the cluster examining gaze behaviour. To do so, we amended our origi- with two objects is to the left of the single object), the latter nal task (Segen et al., 2020) in a variety of ways to overcome manipulation required a fine-grained spatial representation the limitations of the earlier study. First, we reduced the size of the exact positions of the objects as the overall relation- of the perspective shift between encoding and test which ships between the clusters was maintained. allowed us to present the same landmarks during learning Consistent with previous research, we found that older and test, ensuring that participants could use the informa- adults had greater difficulty with the task than younger adults tion they encoded during learning to solve the task at test. (Hartley et al., 2007; Hilton et al., 2020; Montefinese et al., Decreasing the size of the perspective shift also made the 2015; Muffato et al., 2019). Diffusion modelling showed that task easier (Hegarty & Waller, 2004; Montofinese et al., older adults not only had greater difficulty in extracting use- 2015; Segen et al., 2020; Muffato et al., 2019; Hilton et al., ful information from the stimuli but that they also adopted a 2020). Task difficulty was further reduced by including only more conservative response strategy, i.e. they accumulated the condition in which two object clusters were swapped more information before reaching a decision. with each other. Reducing task difficulty aimed at avoiding Furthermore, the analysis of gaze data in Segen et al. floor level performance in older adults, which would allow (2020) revealed that older adults attended to a larger propor- us to rule out that potential differences in gaze behaviour tion of the scenes compared to younger adults. We proposed across groups are caused by participants’ inability to carry two potential explanations for this. First, differences in gaze out the task. behaviour may reflect differences in encoding strategies with Generally, we predict a decline in performance in older older adults encoding object locations relative to the land- adults consistent with age-related place recognition deficits marks available in the room (windows, door, etc.), whilst (Hartley et al., 2007). Responding after a perspective shift young adults focus on the local arrangement of objects and requires additional and demanding mental manipulations of on encoding the spatial relationships among them. The dif- the stored representations (e.g., mentally rotating the new ferences in encoding strategies may reflect a shift towards or the stored representation to match the other, imagining categorical spatial representations in older adults, driven moving around the array; Hegarty & Waller, 2004; Holmes by age-related hippocampal neurodegeneration (Antonova et al., 2018; King et al., 2002). Therefore, we expect that the et al., 2009; Meulenbroek et al., 2004; Moffat et al., 2007). introduction of the perspective shift would impair perfor- Second, older adults may have difficulties in focusing mance in both groups. However, we predict a larger decrease on the task-relevant information as they become distracted in performance in older adults who seem to have difficul- by salient features within the environment. This is in line ties with initiating those mental manipulations as reflected with the attention inhibition deficit in ageing reported in in past findings documenting larger impairments with the past studies (e.g., Hasher & Zacks, 1988). According to this introduction rather than the increase of the perspective shift account, older adults exhibit top-down control difficulties, (Montofinesse et al., 2015; Muffato et al., 2019; Hilton et al., with attention orienting being more affected by stimulus 2020; Segen et al., 2020). properties rather than the task at hand (Olk & Kingstone, To investigate the role of landmarks in encoding 2015; West, 1996). Lastly, older adults may have difficulties strategies and performance, we included trials in which in selecting the appropriate information required to solve the landmarks (in the form of posters on the walls) were: (1) task. This is consistent with our findings that older adults unique and could be used to encode object locations, (2) have difficulties in extracting useful information from the identical and thus uninformative or (3) absent from the stimuli (Segen et al., 2020). scene. Varying the availability and utility of room-based In our earlier study (Segen et al., 2020), we could not landmarks allowed us to test whether age-related differ - distinguish between these explanations for several reasons. ences in gaze behaviour during spatial encoding were due First, we did not systematically manipulate the availability to older adults encoding object positions by relating them 1 3 406 Psychological Research (2022) 86:404–420 to the landmarks or to older adults having difficulties in Virtual environment selecting and/or focusing on task-relevant information. Since this part of the study is largely exploratory, we The virtual environment was designed with Adobe 3DS have formulated a series of predictions about results that Max 2018 and depicted a 13.5  m × 14.6  m rectangular we would expect to find depending on how older adults room. The room contained 6 identical objects; pink vases use additional landmarks during encoding of object loca- on metal stands that were arranged in three clusters of 1, 2 tions. Given that the task can be solved either by focusing and 3 objects in the centre of the room (see Fig. 1). In the on the local arrangement of objects or by relating object No Landmarks condition, the walls contained no additional positions to landmarks, we should not necessarily expect cues, in the Uninformative Landmarks condition eight iden- age-related differences in performance if older adults sim- tical posters of the Tower Bridge were presented, two on ply shift towards a particular encoding strategy depending each wall. Finally, in the Informative Landmarks condition on which information is available. However, if older adults eight unique posters were presented, again two on each wall. select an encoding strategy that depends on the availability These posters consisted of highly familiar and recognisable of landmarks as suggested by our previous research (Segen landmarks (Hamburger & Röser, 2014): the Leaning Tower et al., 2020), we expect them to perform better when land- of Pisa, Stonehenge, the Statue of Liberty, the Golden Gate marks are informative than uninformative. Finally, if older Bridge, the Eiffel Tower, the White House, the Big Ben, and adults have difficulties focusing on task-relevant informa - the Great Wall of China. tion as a result of an attention inhibition deficit (Hasher & The experimental stimuli were renderings of the environ- Zacks, 1988), and are therefore distracted by the presence ment with a 70° horizontal field of view (FOV) with a 15% of landmarks, we predict worse performance when land- downward shift in the vertical FOV, yielding an asymmet- marks are available (either informative or uninformative) ric viewing frustum to simulate human vision. The virtual than when they are not. cameras from which the static images of the scenes were In terms of gaze behaviour, if older adults rely more on rendered were arranged on a circle (radius of 6.7 m) at 30° landmarks as part of their encoding strategy, compared to intervals, providing 12 possible camera positions and the their younger counterparts, we expect them to spend more object clusters were arranged in six unique layouts within time gazing at informative landmarks than uninformative the room (Fig. 1b-d). Six of those camera positions were landmarks. If, however, older adults are distracted by the used in the learning phase and in the 0° perspective shift landmarks, we expected them to show similar gaze behav- condition. The remaining 6 viewpoints were used in the iour in conditions with informative and uninformative test phase in the 30° perspective shift condition. Stimuli landmarks. were presented as static images on a desktop computer with OpenSesame 3.1.7 (Mathôt et al., 2012) and a standard com- puter keyboard was used to collect responses. Method Eye‑tracking Participants Eye movements were recorded using an Eyelink II (SR Research) head-mounted eye tracker at a rate of 500 Hz. Twenty-eight young (mean age = 21.00 years, SD = 2.27; age Calibrations were performed at least three times and drift range = 18–27 years; 15 females and 13 males) and 32 older correction was performed before each trial. The experiment adults aged 60 years and over (mean age = 68.80, SD = 6.34, was presented on a 102 cm screen (diagonal) with an aspect age range = 60–85; 17 females and 15 males) took part in ratio of 16:9 and a resolution of 1920 × 1080 pixels. Par- this study. Participants were recruited either through the ticipants were seated 100 cm from the monitor resulting in participant recruitment system of Bournemouth University a physical horizontal FOV of 47.9° and 28° vertical FOV. or through opportunity sampling in the community. Older adults received monetary compensation for their time whilst Procedure younger participants received course credits. Participants were screened for mild cognitive impairment using the Each experimental trial started with a fixation cross and a Montreal Cognitive Assessment (MoCA; Nasreddine et al., scrambled stimuli mask presented for 1500 ms (Fig. 1a). In 2005). Based on a threshold score of 23, no participants the learning phase, participants were presented with a ren- were excluded (Luis et al., 2009; Waldron-Perrine & Axel- dering of one of the 6 unique configurations of the target rod, 2012). All participants gave their written informed con- objects from one of the six possible viewpoints for 12 s. sent in accordance with the Declaration of Helsinki (World After this learning phase, participants were again presented Medical Association 2013). with a fixation cross and a scrambled stimuli mask for 1 3 Psychological Research (2022) 86:404–420 407 Fig. 1 a Experimental protocol; b, c and d Virtual environment and arrow) and swapping of the two object clusters (Black arrow) in stimuli for the experimental task, Blue and Green cameras repre- Swap trials (b, c). The middle and right panels show the two corre- sent the possible virtual cameras positions for the Learning and Test sponding snapshots for the learning and test phases, respectively. In b phase, respectively. Examples of possible object cluster layouts are and d there is a 30° perspective shift, to the left and right respectively. shown in b (No Landmarks), c (Uninformative Landmarks) and d In c there is no perspective shift. The black arrows in the right panel (Informative Landmarks). The left panel shows a survey perspective (b, c) indicate which clusters were swapped on the test stimuli of the example trials, indicating the rotation of the camera (Orange 1 3 408 Psychological Research (2022) 86:404–420 1500 ms. Then, in the test phase they were presented with a response time distribution. Prior to transforming, response rendering of the room either from the same viewpoint (50% times below 200 ms and over 20,000 ms were removed. of trials, Fig.  1c) or a different viewpoint that was offset by 30° from the study viewpoint (Fig. 1b, d). Participants were asked to respond by pressing the x or m keys on the Results keyboard as to whether the target objects were in the same locations as during the training phase or not. In 50% of the Accuracy trials, the target objects remained in the same locations (Same, Fig. 1c) and in the other 50% of the trials, two of the Accuracy estimates were obtained for each participant with three object clusters swapped locations (Swap, Fig. 1b, c). Age Group, Perspective Shift, Landmark Type and Manipu- As a result, chance level performance for this task was 50%. lation as fixed factors and a random by-subject and by-item The experiment consisted of 144 experimental trials that intercept. Coefficients, standard errors and z -values (Table 1) were preceded by 6 practice trials. The entire study took indicate that Perspective Shift and Manipulation affected around 90 min to complete and participants were allowed to performance. Specifically, accuracy decreased with the take breaks when they wished. introduction of a 30° Perspective Shift (Fig. 2a) and in the Swap condition (Fig. 2b). In addition, there was an inter- action between Perspective Shift and Manipulation with a Design greater decline in performance in the No Change condition compared to the Swap condition following a 30° Perspective The experiment followed a mixed 2 (Age Group: young vs. Shift (Fig.  2c). Finally, we found a three-way interaction older adults) × 2 (Manipulation: Same, Swap, Fig. 1b,c and between Perspective Shift, Manipulation and Age Group d) × 2 (Perspective Shift: 0°, 30°) × 3 (Landmark Type: No with older adults showing a larger decline in performance Landmarks, Uninformative, Informative) design with Manip- than younger adults in the No Change condition when a ulation, Perspective Shift and Landmark Type manipulated 30° Perspective Shift was introduced, whilst displaying an within participants and Age Group between. increase in performance in the Swap condition when a 30° Perspective Shift was introduced (Fig. 2d). Effect plots for Data Analysis significant main effects and interactions are reported in the Supplementary Materials. Data from one older participant were excluded from all analyses due to chance level performance in the 0° Perspec- Response Time tive Shift condition. The remaining data were analysed with linear mixed-effects models (LME) using LME4 (Bates As with accuracy, response time estimates were obtained for et al., 2018) in R (R Core Team, 2013). Specifically, accu- each participant with Age Group, Perspective Shift, Landmark racy was analysed using generalized linear mixed-effects Type and Manipulation as fixed factors and a random by-sub- (GLME) models with the glmer function from LME4 pack- ject and by-item intercept with a random slope for Manipu- age. The following contrasts were used in all (G)LMEs con- lation across participants. Coefficients, standard errors and ducted: Age Group (Younger adults/Older adults), Perspec- t-values (Table 2) show that Age Group, Perspective Shift, tive shift (0°/30°) and Manipulation (No Change/Swap) were Landmark Type and Manipulation were all reliable predic- coded using effect coding. This coding scheme compares tors of response time. Specifically, we found that older adults the effect of a variable (i.e. Age Group) on performance were slower to respond compared to younger adults (Fig. 3a), averaged across all levels of other variables (i.e. Perspec- and that response times increased with the introduction of tive Shift and Manipulation). Landmark Type was coded a Perspective Shift (Fig.  3b). In addition, response times using treatment coding. Since we were interested in examin- were longer with Informative than Uninformative Landmark ing the difference between Informative and Uninformative Type (Fig. 3c) and in the Swap condition compared to the No Landmarks and the difference between No Landmarks and Change condition (Fig. 3d). We also found a significant inter - Uninformative Landmarks, we used the Uninfomative Land- action between Age Group and Manipulation with a smaller mark as the baseline. As a result, all of the effects for other increase in response times in the Swap condition in older than factors are calculated with reference to the performance in younger adults (Fig. 3e). There was also a Perspective Shift and the Uninformative Landmark, rather than the average of per- Manipulation interaction with a smaller increase in response formance for all levels of Landmark Type. For the response times in the Swap condition than the No Change condition time analysis, we included only the correct trials and we log- with the introduction of the Perspective Shift (Fig. 3f). We also transformed response times following the recommendations found an interaction between Landmark Type and Manipula- of Baayen et al. (2008) for dealing with the skewness of the tion with a smaller increase in response times between the No 1 3 Psychological Research (2022) 86:404–420 409 Table 1 Coefficients from Accuracy GLME analysis Predictors Accuracy Coefficients Std. Error z-value (Intercept) 2.023 0.262 7.724 Age Group (Old) 0.145 0.112 1.293 Perspective Shift (30°) − 0.635 0.079 − 8.049 Landmark Type (Informative) − 0.122 0.347 − 0.350 Landmark Type (No Landmarks) 0.066 0.350 0.189 Manipulation (Swap) − 1.316 0.086 − 15.216 Age Group (Old): Perspective Shift (30°) − 0.104 0.071 − 1.468 Age Group (Old): Landmark Type (Informative) − 0.063 0.095 − 0.659 Age Group (Old): Landmark Type (No Landmarks) − 0.138 0.105 − 1.314 Perspective Shift (30°): Landmark Type (Informative) 0.176 0.106 1.659 Perspective Shift (30°): Landmark Type (No Landmarks) − 0.037 0.116 − 0.319 Age Group (Old): Manipulation (Swap) 0.063 0.071 0.887 Perspective Shift (30°): Manipulation (Swap) 0.414 0.077 5.387 Landmark Type (Informative): Manipulation (Swap) 0.212 0.115 1.846 Landmark Type (No Landmarks): Manipulation (Swap) − 0.082 0.125 − 0.656 Age Group (Old): Perspective Shift (30°): Landmark Type (Informative) 0.097 0.095 1.020 Age Group (Old): Perspective Shift (30°): Landmark Type (No Landmarks) 0.137 0.105 1.303 Age Group (Old): Perspective Shift (30°): Manipulation (Swap) 0.240 0.071 3.399 Age Group (Old): Landmark Type (Informative): Manipulation (Swap) 0.049 0.095 0.514 Age Group (Old): Landmark Type (No Landmarks): Manipulation (Swap) 0.054 0.105 0.512 Perspective Shift (30°): Landmark Type (Informative): Manipulation (Swap) 0.060 0.103 0.584 Perspective Shift (30°): Landmark Type (No Landmarks): Manipulation (Swap) 0.155 0.114 1.364 Age Group (Old): Perspective Shift (30°): Landmark Type (Informative): Manipulation (Swap) − 0.122 0.095 − 1.277 Age Group (Old): Perspective Shift (30°): Landmark Type (No Landmarks): Manipulation (Swap) − 0.201 0.105 − 1.916 Significant z values (|z|≥ 1.96) in bold Fig. 2 Bar plots of accuracy values for a significant main effect of a Perspective Shift, b Manipulation, and sig- nificant interactions c between Manipulation and Perspective Shift and d Interaction between Age Group, Manipulation and Perspective Shift with a mean (solid line) and 95% CIs (grey shaded area) with individual data points and violin plots behind 1 3 410 Psychological Research (2022) 86:404–420 Table 2 Coefficients from response time LME analysis Predictors Log transformed response time Estimates Std. Error t-value (Intercept) 7.834 0.041 190.067 Age Group 0.209 0.040 5.248 Perspective Shift (30°) 0.130 0.015 8.459 Landmark Type (Informative) 0.058 0.020 2.942 Landmark Type (No Landmarks) − 0.013 0.020 − 0.640 Manipulation (Swap) 0.133 0.011 12.386 Age Group: Perspective Shift (30°) 0.006 0.014 0.451 Age Group: Landmark Type (Informative) 0.019 0.013 1.470 Age Group: Landmark Type (No Landmarks) − 0.019 0.013 − 1.443 Perspective Shift (30°): Landmark Type (Informative) 0.012 0.015 0.813 Perspective Shift (30°): Landmark Type (No Landmarks) − 0.000 0.015 − 0.007 Age Group: Manipulation (Swap) − 0.032 0.009 − 3.474 Perspective Shift (30°): Manipulation (Swap) − 0.077 0.010 − 7.542 Landmark Type (Informative): Manipulation (Swap) − 0.034 0.015 − 2.259 Landmark Type (No Landmarks): Manipulation (Swap) 0.010 0.015 0.654 Age Group: Perspective Shift (30): Landmark Type (Informative) − 0.003 0.013 − 0.239 Age Group: Perspective Shift (30°): Landmark Type (No Landmarks) − 0.013 0.013 − 1.012 Age Group: Perspective Shift (30°): Manipulation (Swap) − 0.018 0.009 − 1.960 Age Group: Landmark Type (Informative): Manipulation (Swap) − 0.008 0.013 − 0.596 Age Age Group: Landmark Type (No Landmarks): Manipulation (Swap) − 0.024 0.013 − 1.847 Perspective Shift (30°): Landmark Type (Informative): Manipulation (Swap) 0.019 0.014 1.312 Perspective Shift (30°): Landmark Type (No Landmarks): Manipulation (Swap) 0.002 0.014 0.162 Age Group: Perspective Shift (30°): Landmark Type (Informative): Manipulation (Swap) 0.005 0.013 0.406 Age Group: Perspective Shift (30°): Landmark Type (No Landmarks): Manipulation (Swap) − 0.004 0.013 − 0.289 Significant t values (|t|≥ 1.96) in bold Change and the Swap condition in the Informative Landmark Theory evaluates sensitivity and response bias in situa- Type (Fig. 3g) compared to Uninformative Landmark Type tions that require decision making under uncertainty. It condition. Finally, we found a three-way interaction between is applied when a binary decision about the presence or Age Group, Perspective Shift and Manipulation, with the Age absence of a signal is made, comparing the response with Group and Perspective Shift interactions showing a different the actual presence/absence of the signal. With Signal trend across No Change and Swap Manipulation. Specifically, Detection Theory, the formula c = -0.5[z(hit rate) + z(false there was a larger increase in response times in older adults alarm rate) is used to compute response bias, where hit than young adults, in the No Change condition with the intro- rate and false alarm rates refer to trials in which the sig- duction of the Perspective Shift (Fig. 3). Whilst in the Swap nal was correctly or incorrectly, respectively, reported as condition, the increase in response times in older adults was present. smaller when a Perspective Shift was introduced compared Overall, there was a positive response bias showing to young adults. Effect plots for significant main effects and that participants were more likely to respond that nothing interactions are reported in the Supplementary Materials. has changed than to respond that something had changed (Fig.  4). LMM analysis (Table  3) with Age Group, Per- spective Shift and Landmark Type as fixed factors and by- Response Bias subject intercept with a random slope for Perspective Shift, indicated that the introduction of a Perspective Shift led To examine if participants displayed a response bias, we to a decrease in response bias, which was larger in older carried out an analysis based on Signal Detection Theory adults than in younger adults. Furthermore, when a Per- (Harvey, 1992; Macmillan & Creelman, 1991) using the spective Shift was introduced, the response bias decreased sdt.rmcs (Todorova, 2017) package in R. Signal Detection more in the No Landmarks and the Informative Landmarks 1 3 Psychological Research (2022) 86:404–420 411 Fig. 3 Bar plots of Response Times values for significant main effects only for Landmark Type (Informative): Manipulation (Swap)) H: Age of A: Age Group B: Perspective Shift C: Landmark Type (signifi- Group, Manipulation and Perspective Shift with mean (solid line) and cant only for the Informative Landmark Type) D: Manipulation and 95% CIs (grey shaded area) with individual data points and violin interactions between E: Age Group and Manipulation F: Perspective plots behind Shift and Manipulation G: Landmark Type and Condition (significant 1 3 412 Psychological Research (2022) 86:404–420 Fig. 4 Bar plots for Response Bias as a function of Age Group, Landmark Type and Per- spective Shift with mean (solid line) and 95% CIs (grey shaded area) with individual data points and violin plots behind. Stars indicate response bias signifi- cantly different from 0 (1 star [p < 0.05], 2 stars [p < 0.01] and 3 stars [p < 0.001]) Table 3 Coefficients from Response Bias LME analysis Predictors Response bias (c) Estimates Std. Error t-value (Intercept) 0.437 0.033 13.043 Age Group (Older Adults) − 0.047 0.033 − 1.403 Perspective Shift (30°) − 0.069 0.029 − 2.384 Landmark Type (Informative) 0.003 0.026 0.097 Landmark Type (No Landmarks) − 0.048 0.026 − 1.826 Age Group: Perspective Shift (30°) − 0.072 0.029 − 2.495 Age Group: Landmark Type (Informative) 0.007 0.026 0.264 Age Group: Landmark Type (No Landmarks) − 0.008 0.026 − 0.306 Perspective Shift (30°): Landmark Type (Informative) − 0.052 0.026 − 1.978 Perspective Shift (30°): Landmark Type (No Landmarks) − 0.058 0.026 − 2.201 Age Group: Perspective Shift (30°): Landmark Type (Informative) 0.049 0.026 1.845 Age Group: Perspective Shift (30°): Landmark Type (No Landmarks) 0.028 0.026 1.043 Significant t values (|t|≥ 1.96) in bold conditions compared to the Uninformative Landmarks con- Gaze analysis dition (Table 4). Fixations and saccades were identified using the SR Research algorithms and were pre-processed as follows: First, we removed fixations that contained a blink, fell 1 3 Psychological Research (2022) 86:404–420 413 Table 4 Coefficients from Dwell Time on the top IA LME analysis predictors of Dwell Time on the top IA. Specifically, we found that compared to the Uninformative Landmarks Predictors Dwell Time on Landmarks condition that was used as a baseline, there was a reduc- Estimates Std. Error t-value tion in Dwell Time on the top IA in the No Landmarks and an increase in Dwell Time in the Informative Landmarks (Intercept) 13.054 1.503 8.684 condition. We also found that older adults spent more time Age Group (Older Adults) 2.99/ 1.457 2.058 looking at the top IA compared to younger adults. In addi- Landmark Type (No Land- − 8.108 0.644 − 12.600 marks) tion, there was a Landmark Type and Age Group interaction Landmark Type (Informative) 9.540 0.644 14.826 whereby older adults’ Dwell Time on Landmarks decreased Age Group (Older Adults): − 1.804 0.375 − 4.812 more than that of younger adults’ in the No Landmarks con- Landmark Type (No Land- dition compared to Uninformative Landmarks condition and marks) showed a larger increase in the Informative Landmarks con- Age Group (Older Adults): 1.171 0.375 3.124 dition compared to the Uninformative Landmarks condition. Landmark Type (Informa- A Dwell Time analysis on the top IA at test produced similar tive) results to those of the learning phase, with the exception Significant t values (|t|≥ 1.96) in bold that the increase in Dwell Time in older adults and the Age Group by Landmark Type (No Landmarks) interaction were not significant. Results from this analysis are presented in outside of the screen boundaries or were shorter than 80 ms the Supplementary Materials. or longer than 1000 ms (Inhoff & Radach, 1998; Nuthmann, 2017). Finally, we removed saccades with maximum ampli- Relationship between Gaze and Performance tudes (41.35°va) or velocities (1500°/s) larger than it should be possible based on the distance of the participant from the Dwell time on the top IA was not related to performance screen and the screen size. across any of the three Landmark Type conditions (Fig. 6), The primary aim of the gaze analysis was to investigate suggesting that the task could be solved either by using age differences in encoding strategies and was therefore Landmarks (when they are available) or by focusing primar- mainly focused on the analysis of gaze during the encoding ily on the objects. Thus, the differences in gaze behaviour phase. Analysis of differences in basic saccade and fixation reported here are likely to represent differences in encoding parameters between young and older adults showed that strategy preferences that change with age. during the 12 s encoding period, older adults made shorter and more frequent fixations as well as more frequent sac- Gaze behaviour across trials cades. The results are reported in detail in the supplementary materials. We also investigated if gaze behaviour changes across time by correlating Dwell Time on landmarks with trial older Gaze on landmarks for younger and older participants in the No Landmark, Uninformative and Informative Landmark conditions. We As we were primarily interested in age-related differences found that across both younger and older adults, Dwell Time in gaze as a function of Landmark Type, we split stimuli remained consistent in the No Landmark condition through- into two interest areas (See Fig. 5) and compared the per- out the experiment (Young: r = 0.011, p = 0.895, Older: centage of Dwell Time on the top interest area (IA) where r = − 0.09, p = 0.279). In the Uninformative Landmark con- Landmarks were located when available vs. the bottom area dition, older adults spent less time fixating on landmarks where the objects were located. To do so, we computed the over the course of the experiment (r = − 0.18, p = 0.032), total dwell time for each trial by adding up the duration of whilst younger adults’ gaze (r = − 0.05, p = 0.543) remained all fixations in the trial. Next, we calculated the proportion unchanged. In the Informative Landmark condition, an oppo- of dwell time that was spent fixating in the top IA. This site pattern of results was found with younger adults spend- approach allowed us to specifically focus on age-related ing less time fixating on landmarks (r = − 0.20, p = 0.018) differences in the use of room-based Landmarks during with older adults’ gaze remaining unchanged (r = − 0.09, encoding with the increased Dwell Time on the upper part p = 0.266). of the stimuli largely reflecting gaze on Landmarks (when available). Consistency in gaze between learning and test LME analysis with Age Group and Landmark Type as fixed factors and a by-subject and by-item random intercept Finally, we examined if participants showed similar gaze showed that Landmark Type and Age Group were reliable behaviour during learning and test. To do so, we correlated the 1 3 414 Psychological Research (2022) 86:404–420 Fig. 5 Heatmaps representing number of fixations as a func- tion of age group and landmark type Dwell Time on the top IA across different Landmark Types at Discussion learning and test. We found strong positive correlations across all Landmark Types (No Landmarks: R2 = 0.67, p < 0.001; In the present study, we used eye-tracking to investi- Uninformative: R2 = 0.88, p < 0.001; Informative: R2 = 0.94, gate age-related differences in visual encoding strategies p < 0.001). Those correlations suggest that participants are employed for memorizing the locations of objects in a highly consistent in which stimulus features they gaze at dur- room. To do so, we explored if participants were able to ing encoding and test. 1 3 Psychological Research (2022) 86:404–420 415 of previous studies (Muffato et al., 2019; Montofinese et al., 2015; Segen et al., 2020). In addition, the scene at test could differ from the encoded only in terms of a change in the categorical relationship between objects. That is, in contrast to Segen et al. (2020), in the current study no changes in fine-grained spatial relationships between objects occurred. That the easier task may be responsible for the lack of age- related deficits in task accuracy is in line with cognitive age- ing research reporting greater age-related differences in per - formance with increasing task difficulty (Angel et al., 2016; Earles et al., 2004; Verhaeghen et al., 2006). The lack of age-related performance accuracy differences in less demanding tasks can be explained by the compensa- tion-related utilization of neural circuits hypothesis (Reuter- Lorenz & Cappell, 2008). This hypothesis posits that under low task demands older adults can perform the tasks as well as young adults, supported by increased neural activations. Fig. 6 Scatter Plot between Dwell Time on the top IA and Accuracy However, when task demands increase, older adults’ cogni- as a function of Landmark Type with regression line and CI (shaded tive limits are reached resulting in performance declines that area) are typically accompanied by a reduction in activation in the relevant neural networks (Morcom & Rugg, 2007; Angel identify whether a spatial scene has changed following a et al., 2016). Thus, it is plausible that due to the relatively perspective shift between encoding and test. The 30° per- low task-demands in the current study, which are reflected spective shift was introduced to ensure that participants in high performance across both age groups, older adults relied on spatial representations instead of solving the were able to carry out the task just as accurately as younger task by matching the visual image with a stored visual participants. snapshot from encoding (Nardini et al., 2009). To inves- Consistent with our predictions, we found declines in tigate the effect of landmarks on encoding strategies, we accuracy in both age groups that were accompanied by an also manipulated the availability and informative value of increase in response times when a perspective shift was landmarks within the environment. introduced. This reduction in performance may have been We found that overall, older adults took longer to respond. driven by qualitative differences between trials that involved This increase in response times is consistent with findings a perspective shift and those that did not. Specifically, with- that are widely reported in the cognitive ageing literature out a perspective shift participant can refer to the represen- (Choice reaction time task: Woods et al., 2015; Memory: tation of the learned scene from memory and use image Hertzog et al., 2003; Language: Ratcliff et al., 2004a, b), and matching to detect changes (Nardini et al., 2009). However, istypically attributed to decrements in speed of processing the introduction of the perspective shift required partici- (Salthous, 1996; Salthouse & Ferrer-Caja, 2003). We also pants to engage in additional cognitive processing related found that the introduction of the perspective shift and the to mental transformation in order to match the perspectives manipulation of object positions led to performance dec- of the stored spatial configuration with the one presented at rements in both age-groups. The availability and informa- test (Hegarty & Waller, 2004). However, it should be noted tiveness of the room-based landmarks did not affect task that the effect of the perspective shift was small, which is accuracy. Importantly, we found that when landmarks were likely due to the relatively small perspective shift that we presented, older participants spent more time than younger introduced. participants looking at the upper part of the display that con- Interestingly, there was a much more nuanced (if any) tained the landmarks. This was particularly the case when decline in accuracy or increase in response time in the Swap the landmarks were informative. compared to the No Change condition when a perspective Contrary to our expectations and previous place recogni- shift was introduced. To explain such findings, we turn to the tion research (Muffato et al., 2019; Hilton et al., 2020; Segen response bias analysis which suggested that the introduction et al., 2020; Harley et al., 2007), there were no age-related of the perspective shift increased the likelihood of partici- differences in accuracy. However, it should be noted that pants responding that the object positions were “different”. we used an easier task than those used in previous studies, Thus, when a perspective shift was introduced in the Swap which could yield fewer problems for older adults. For exam- condition, this led to an increase in the number of correct ple, the perspective shift we introduced was smaller than that responses albeit for the wrong reason. We believe that the 1 3 416 Psychological Research (2022) 86:404–420 increase of “different” responses after a perspective shift to use eye-tracking to investigate age-related differences arises from the salient change in the visual input indicating in spatial encoding strategies and to study if such differ - that “something is different”. However, if participants were ences are driven by the information available within the solely responding to any change in the visual information environment. Firstly, we focused on general gaze param- between encoding and test, we expected them to perform eters and found that older adults made more fixations that below chance level in the No Change condition when a per- were shorter in duration as well as shorter saccades than spective shift was present. Yet, our participants were still young adults. While these results are consistent with those able to perform well in this condition and their performance from a recent study using a similar place recognition task in the Swap condition with perspective shifts was not at the (Hilton et al., 2020), relating these general gaze measures ceiling. This pattern of results demonstrates that participants to encoding strategies is difficult. We thus performed IA were not solely relying on basic visual change detection but analysis which showed that gaze behaviour differed as a were instead using a spatial strategy to perform the task. function of room type. As expected, we found that both Yet, they might have found it hard to inhibit the immediate age groups spent the lowest amount of time looking at the response that the image is “not the same” when the perspec- upper part of the stimuli in the No Landmarks condition tive shift was introduced. The increase in performance in in which there were no images on the walls of the room, older adults with the introduction of the perspective shift in followed by the Uninformative Landmarks condition, in the Swap condition may thus be due to older adults expe- which the images on the walls were all identical, and the riencing even greater difficulty in inhibiting the response Informative Landmarks condition in which each image was that the image is “not the same” when a perspective shift unique. These findings are consistent with results reported was present. Such difficulties are in line with age-related by Livingstone-Lee et al. (2011) who showed that partici- decline in executive functioning, in particular executive pants quickly learned to adapt their gaze distribution in a control (Braver & West, 2008; Schretlen et al., 2000; Treitz virtual Morris water maze task based on the information et al., 2007). that was available in the environment. Importantly, we Overall participants were more likely to make errors in found that compared to younger adults, older adults spent the Swap condition than the No Change condition. To per- more time looking at landmarks in the Uninformative and form the task accurately participants in either condition had Informative Landmarks conditions during encoding. A to bind an object’s identity to its location (Postma et al., similar pattern was observed during the test phase in the 2004; Waller, 2006). Previous research has shown that this is Informative Landmarks condition. a cognitively demanding and error-prone process. For exam- A possible explanation for these age-related-differences ple, in place recognition studies participants were shown in gaze behaviour is that older adults simply look around to be less accurate in detecting that a change has occurred more due to a lack of a systematic encoding strategy. This when two objects swapped places compared to when a previ- can arise as a result of difficulties in selecting task-relevant ously shown object was replaced by a new one (Hilton et al., information (Raptis et al., 2017). Given our results, how- 2020; Muffato et al., 2019). Similar results are reported in ever, it appears unlikely that older adults were randomly visuospatial working memory studies in which participants scanning the environment without a clear encoding strat- were asked to encode positions of abstract objects on a blank egy for several reasons: first, older adults solved the task display. Participants were more likely to make swap errors, as accurately as younger participants, which would not be that is to place objects in the positions that were previously possible without a clear encoding strategy. Second, we found occupied by a different object (Pertzov et al., 2012, 2015). that older adults’ gaze behaviour changed as a function of Thus, the lower performance in the Swap condition can the landmarks used. Specifically, older adults spent signifi- be explained by difficulties with binding objects to their cantly more time looking at the upper part of the stimuli locations, which prevents participants from accumulating when landmarks were present and when these landmarks information signalling that a change has occurred (Hilton were informative, i.e. when they could be used to encode et al., 2020; Muffato et al., 2019). Specifically, in the current the spatial locations of the objects by relating objects to task, the objects within the scene and their general configu- these room-based landmarks. Third, both younger and older ration remained the same between learning and test. The adults adapted their gaze behaviour over the course of the only change introduced in the Swap condition is the posi- experiment such that older adults spent less time fixating on tion that each cluster occupied within that general configura- uninformative landmarks across trials. Younger participants, tion. Therefore, participants needed to remember the specific on the other hand, spent less time fixating on informative locations of each object cluster within that configuration to landmarks across the trial. These changes in gaze behaviour detect that a change has occurred. over time are likely to reflect adaptations of encoding strate- In addition to comparing the behavioural performance gies with older adults learning to inhibit attending to unin- of older and younger adults, another aim of this study was formative information and younger participants focusing on 1 3 Psychological Research (2022) 86:404–420 417 encoding the relationship between objects even in the pres- enough resources to deal with the task at hand and if they are ence of informative landmarks. directing already limited resources to task-irrelevant infor- Finally, gaze behaviour was highly consistent between mation (Angel et al., 2016; Morcom et al., 2007; Reuter- learning and test, which suggests that participants, both Lorenz & Cappell, 2008). young and older, attended to the same information during The idea that older adults have a greater preference than learning and test. It is possible that low-level properties of young adults towards encoding strategies that incorporate all the stimuli (i.e. colour, intensity and orientation) contributed available landmarks is consistent with results from research to such similarities in gaze behaviour through bottom-up that employs diffusion modelling. Several studies document control of attention (Itti, 2005), as similar visual information an age-related shift towards a more conservative response was presented at both learning and test. However, given that strategy whereby, compared to young adults, older adults participants performed well on the task and made very few prefer to accumulate more information before making deci- fixations at test, it is unlikely that the consistency between sions (Ratcliff et al., 2006, 2004a, b; Segen et al., 2020; gaze behaviour at learning and test was solely driven by Spaniol et  al., 2006; Thapar et  al., 2003). This explana- bottom-up processes. Instead, we suggest that participants tion is also supported by our findings of longer response relied on the information they encoded at learning to make times in older adults which could be indicative of greater decisions regarding whether or not the objects have moved cautiousness. at test. Together, these results suggest that gaze behaviour, Alternatively, the preference for attending to landmarks in both younger and older adults, represents task and stim- during encoding could be indicative of age-related differ - uli-dependent visual strategies that participants employed ences in spatial encoding strategies. Specifically, older to solve the task. adults’ may be more reliant on an encoding strategy in Age-related differences in gaze behaviour may also be which they relate the positions of objects to landmarks, driven by older adults being distracted by salient, but task- while younger participants focus on the local arrangement irrelevant landmarks as a result of attention inhibition defi- of objects and encode the spatial relationships between cits (Hasher & Zack, 1988; Healey et al., 2008, 2013). This them. This interpretation is in line with our findings that account is partly supported by our findings as older adults older adults spent more time than younger adults looking spent more time than younger adults gazing at the uninform- at the landmarks during encoding, especially when these ative landmarks. Notably, however, this did not affect their were informative. The differences in encoding strategies may performance and can be explained by the relatively long represent an age-related shift towards the use of a categori- encoding times that allowed participants to encode adequate cal encoding strategy whereby participants bind an object task-relevant information even if they were briefly distracted. to the nearest cue/landmark without the need to encode the An alternative explanation as to why older adults attended exact metric relationship between the two. This shift may to uninformative landmarks (i.e. task-irrelevant informa- arise from difficulties in forming precise spatial representa- tion), is a more general age-related shift in the way they tions. For example, previous visuospatial working memory approach cognitive tasks. Zimmerman et al. (2011) sug- research has shown that older adults were less precise in esti- gested that older adults tend to implicitly encode all of the mating previous locations of objects compared to younger available information, regardless of its immediate utility. adults, despite positioning the objects in the correct region This is consistent with evidence showing that the inability to of the stimuli (Nilakantan et al., 2018; Pertzov et al., 2015). inhibit attention sometimes comes with benefits. Kim et al. Furthermore, in navigation, older adults show greater pref- (2007), for example, have shown that older adults display erence towards the use of beacon strategies (Wiener et al., greater priming benefits when distractors on a previous task 2013). Such strategies involve coarse categorical represen- were used as primes in a problem-solving task. It is pos- tations of locations in relation to environmental beacons or sible that the shift towards encoding irrelevant, as well as landmarks and may be preferred by older adults due to dif- relevant information, stems from greater experience with ficulties in formulating more precise representations. real-world environments in which apparently task-irrelevant Lastly, we did not find any relation between gaze behav - information often becomes relevant in the future (Kim et al., iour and performance. This is not surprising as we found 2007; Zimmerman et al., 2011). For example, remembering similar performance across different room types and across extra landmarks in the environment could help to distinguish both age groups despite the presence of gaze differences. similar environments from each other. Such implicit shifts in These results indicate that the current task can be solved encoding strategies may explain why older adults spent more equally well by focusing on objects and by relating the time looking at extra information even if this information objects to landmarks (if they are available), with older adults is not strictly necessary for solving the task at hand. How- showing a preference towards the latter. In addition, the lack ever, such strategy shifts could lead to performance deficits of correlation between gaze and performance is consistent in cognitively taxing situations, if older adults do not have with our previous findings showing that the Swap condition 1 3 418 Psychological Research (2022) 86:404–420 Neural correlates of successful memory retrieval in aging: do could be solved either by looking around more or by having executive functioning and task difficulty matter? Brain Research, more focused gaze (Segen et al., 2020) outlining that coarse 1631, 53–71. https:// doi. org/ 10. 1016/j. brain res. 2015. 10. 009 spatial representations can be formed using a wider range of Antonova, E., Parslow, D., Brammer, M., Dawson, G. R., Jackson, S. encoding strategies and the available information. H. D., & Morris, R. G. (2009). Age-related neural activity during allocentric spatial memory. Memory, 17(2), 125–143. https://doi. To summarise, our results suggest that under specific con- org/ 10. 1080/ 09658 21080 20773 48 ditions such as the presence of a relatively small perspective Baayen, R. H., Davidson, D. J., & Bates, D. M. (2008). Mixed-effects shift and the introduction of categorical changes within the modeling with crossed random effects for subjects and items. scene, spatial memory is resistant to age-related changes Journal of Memory and Language, 59(4), 390–412. https:// doi. org/ 10. 1016/j. jml. 2007. 12. 005 as older adults perform the task as well as younger partici- Braver, T. S., & West, R. (2008). Working memory, executive con- pants. Furthermore, we report an age-related shift in visual trol, and aging. In F. I. M. Craik and T. A. Salthouse (Eds.), The encoding strategy. Although we cannot completely rule out handbook of aging and cognition (p. 311–372). Psychology Press that these changes in gaze behaviour are driven by inhibi- Earles, J. L., Kersten, A. W., Berlin Mas, B., & Miccio, D. M. (2004). Aging and memory for self-performed tasks: effects of task dif- tory control mechanisms, it seems highly plausible that older ficulty and time pressure. The Journals of Gerontology Series adults, who might be more distracted by the uninformative B: Psychological Sciences and Social Sciences, 59(6), 285–293. landmarks, employ an encoding strategy that relies on pro- https:// doi. org/ 10. 1093/ geronb/ 59.6. P285 cessing the categorical relationships between objects and Hamburger, K., & Röser, F. (2014). The role of landmark modality and familiarity in human wayfinding. Swiss Journal of Psychology. room-based landmarks rather than forming fine-grain spatial https:// doi. org/ 10. 1024/ 1421- 0185/ a0001 39 representations. Hartley, T., Bird, C. M., Chan, D., Cipolotti, L., Husain, M., Varga- Khadem, F., & Burgess, N. (2007). The hippocampus is required Supplementary Information The online version contains supplemen- for short-term topographical memory in humans. Hippocampus, tary material available at https://doi. or g/10. 1007/ s00426- 021- 01495-5 . 17(1), 34–48. https:// doi. org/ 10. 1002/ hipo. 20240 Hasher, L., & Zacks, R. T. (1988). Working Memory, Comprehension, and Aging: A Review and a New View. Psychology of Learning Compliance with ethical standards and Motivation - Advances in Research and Theory, 22, 193–225. https:// doi. org/ 10. 1016/ S0079- 7421(08) 60041-9 Conflict of interest The authors declare that the research was con- Healey, M. K., Campbell, K. L., & Hasher, L. (2008). Cognitive aging ducted in the absence of any commercial or financial relationships that and increased distractibility: Costs and potential benefits. Progress could be construed as a potential conflict of interest. in Brain Research, 169, 353–363. https://doi. or g/10. 1016/ S0079- 6123(07) 00022-2 Ethical standards This study was carried out in accordance with the Healey, M. K., Hasher, L., & Campbell, K. L. (2013). The role of recommendations of the Research Ethics Code of Practice, Science, suppression in resolving interference: evidence for an age-related Technology and Health Research Ethics Panel at Bournemouth Univer- deficit. Psychology and Aging, 28(3), 721–728. https:// doi. org/ sity with written informed consent from all subjects and has therefore 10. 1037/ a0033 003 been performed in accordance with the ethical standards laid down in Hegarty, M., & Waller, D. (2004). A dissociation between mental rota- the 1964 Declaration of Helsinki and its later amendments. tion and perspective-taking spatial abilities. Intelligence, 32(2), 175–191. https:// doi. org/ 10. 1016/j. intell. 2003. 12. 001 Open practices statement The datasets generated during and/or Hertzog, C., Dixon, R. A., Hultsch, D. F., & MacDonald, S. W. (2003). analysed during the current study are available in the Open Science Latent change models of adult cognition: are changes in process- Framework repository, https://osf .io/ v4m we/. This experiment was not ing speed and working memory associated with changes in epi- pre-registered. sodic memory? Psychology Aging, 18(4), 755. https://doi. or g/10. 1037/ 0882- 7974. 18.4. 755 Hilton, C., Muffato, V., Slattery, T. J., Miellet, S., & Wiener, J. (2020). Open Access This article is licensed under a Creative Commons Attri- Differences in encoding strategy as a potential explanation for bution 4.0 International License, which permits use, sharing, adapta- age-related decline in place recognition ability. Frontiers in Psy- tion, distribution and reproduction in any medium or format, as long chology, 11, 2182. https:// doi. org/ 10. 3389/ fpsyg. 2020. 02182 as you give appropriate credit to the original author(s) and the source, Holmes, C. A., Newcombe, N. S., & Shipley, T. F. (2018). Move to provide a link to the Creative Commons licence, and indicate if changes learn: Integrating spatial information from multiple viewpoints. were made. The images or other third party material in this article are Cognition, 178, 7–25. https:// doi. org/ 10. 1016/j. cogni tion. 2018. included in the article’s Creative Commons licence, unless indicated 05. 003 otherwise in a credit line to the material. If material is not included in Inhoff, A. W., & Radach, R. (1998). Definition and computation of the article’s Creative Commons licence and your intended use is not oculomotor measures in the study of cognitive processes. In G. permitted by statutory regulation or exceeds the permitted use, you will Underwood (Ed.), Eye guidance in reading and scene perception need to obtain permission directly from the copyright holder. To view a (pp. 29–53). Elsevier. copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . Kelly, D. J., Miellet, S., & Caldara, R. (2010). Culture shapes eye movements for visually homogeneous objects. Frontiers in Psy- chology, 1, 1–7. https:// doi. org/ 10. 3389/ fpsyg. 2010. 00006 Kim, S., Hasher, L., & Zacks, R. T. (2007). Aging and a benefit of References distractibility. Psychonomic Bulletin and Review, 14(2), 301–305. https:// doi. org/ 10. 3758/ BF031 94068 Angel, L., Bastin, C., Genon, S., Salmon, E., Fay, S., Balteau, E., King, J. A., Burgess, N., Hartley, T., Vargha-Khadem, F., & O’Keefe, J. Maquet, P., Luxen, A., Insingrini, M., & Collette, F. (2016). (2002). Human hippocampus and viewpoint dependence in spatial 1 3 Psychological Research (2022) 86:404–420 419 memory. Hippocampus, 12(6), 811–820. https://doi. or g/10. 1002/ Pertzov, Y., Miller, T. D., Gorgoraptis, N., Caine, D., Schott, J. M., But- hipo. 10070 ler, C., & Husain, M. (2013). Binding deficits in memory follow - Livingstone-Lee, S. A., Murchison, S., Zeman, P. M., Gandhi, M., ing medial temporal lobe damage in patients with voltage-gated van Gerven, D., Stewart, L., Livingston, N., & J. and Skelton, R. potassium channel complex antibody-associated limbic encepha- W. . (2011). Simple gaze analysis and special design of a virtual litis. Advance online publication. Brain: A Journal of Neurol- Morris water maze provides a new method for die ff rentiating ego - ogy,  136, 2474–2485.  https://doi.org/ https:// doi. or g/ 10. 1093/ centric and allocentric navigational strategy choice. Behavioural brain/ awt129 brain research, 225(1), 117–125. https:// doi. org/ 10. 1016/j. bbr. Postma, A., Kessels, R. P. C., & van Asselen, M. (2008). How the 2011. 07. 005 brain remembers and forgets where things are: The neurocogni- Luis, C. A., Keegan, A. P., & Mullan, M. (2009). Cross validation of tion of object-location memory. Neuroscience and Biobehavio- the Montreal Cognitive Assessment in community dwelling older ral Reviews, 32(8), 1339–1345. https://doi.org/https://doi. or g/10. adults residing in the Southeastern US. International Journal of 1016/j. neubi orev. 2008. 05. 001 Geriatric Psychiatry, 24(2), 197–201. https:// doi. org/ 10. 1002/ Raptis, G. E., Fidas, C. A., & Avouris, N. M. (2017). On implicit elici- gps. 2101 tation of cognitive strategies using gaze transition entropies in Mathôt, S., Schreij, D., & Theeuwes, J. (2012). OpenSesame: An pattern recognition tasks. In Proceedings of the 2017 CHI Con- open-source, graphical experiment builder for the social sciences. ference Extended Abstracts on Human Factors in Computing Sys- Behavior Research Methods, 44(2), 314–324. https:// doi. org/ 10. tems, 1993–2000. https://doi.org/https://d oi.o rg/1 0.1 145/3 02706 3. 3758/ s13428- 011- 0168-730531 06 Meulenbroek, O., Petersson, K. M., Voermans, N., Weber, B., & Ratcliff, R., Gomez, P., Thapar, A., & McKoon, G. (2004a). A diffusion Fernández, G. (2004). Age differences in neural correlates of route model analysis of the effects of aging in the lexical-decision task. encoding and route recognition. NeuroImage, 22(4), 1503–1514. Psychology and Aging, 19(2), 278–289. https:// doi. org/ 10. 1037/ https:// doi. org/ 10. 1016/j. neuro image. 2004. 04. 0070882- 7974. 19.2. 278 Moffat, S. D., Kennedy, K. M., Rodrigue, K. M., & Raz, N. (2007). Ratcliff, R., McKoon, G., & Gomez, P. (2004b). A diffusion model Extrahippocampal contributions to age differences in human spa- account of the lexical decision task. Psychological Review, 111(1), tial navigation. Cerebral Cortex, 17(6), 1274–1282. https:// doi. 159–182. https:// doi. org/ 10. 1037/ 0033- 295X. 111.1. 159 org/ 10. 1093/ cercor/ bhl036 Ratcliff, R., Thapar, A., & McKoon, G. (2006). Aging, practice, and Montefinese, M., Sulpizio, V., Galati, G., & Committeri, G. (2015). perceptual tasks: a diffusion model analysis. Psychology and Age-related effects on spatial memory across viewpoint changes Aging, 21(2), 353–371. https:// doi. org/ 10. 1037/ 0882- 7974. 21.2. relative to die ff rent reference frames. Psychological Research Psy - 353 chologische Forschung, 79(4), 687–697. https:// doi. org/ 10. 1007/ Reuter-Lorenz, P. A., & Cappell, K. A. (2008). Neurocognitive aging s00426- 014- 0598-9 and the compensation hypothesis. Current Directions in Psycho- Morcom, A. M., Li, J., & Rugg, M. D. (2007). Age effects on the neu- logical Science, 17(3), 177–182. https:// doi. org/ 10. 1111/j. 1467- ral correlates of episodic retrieval: increased cortical recruitment 8721. 2008. 00570.x with matched performance. Cerebral Cortex, 17(11), 2491–2506. Salthouse, T. A. (1996). The processing-speed theory of adult age dif- https:// doi. org/ 10. 1093/ cercor/ bhl155 ferences in cognition. Psychological Review, 103(3), 403. https:// Muffato, V., Hilton, C., Meneghetti, C., De Beni, R., & Wiener, J. M. doi. org/ 10. 1037/ 0033- 295x. 103.3. 403 (2019). Evidence for age-related deficits in object-location bind- Salthouse, T. A., & Ferrer-Caja, E. (2003). What needs to be explained ing during place recognition. Hippocampus, 29(10), 971–979. to account for age-related effects on multiple cognitive variables? https:// doi. org/ 10. 1002/ hipo. 23099 Psychology and Aging, 18(1), 91. https:// doi. org/ 10. 1037/ 0882- Nardini, M., Thomas, R. L., Knowland, V. C., Braddick, O. J., & Atkin-7974. 18.1. 91 son, J. (2009). A viewpoint-independent process for spatial reori- Schretlen, D., Pearlson, G. D., Anthony, J. C., Aylward, E. H., Augus- entation. Cognition, 112(2), 241–248. https:// doi. org/ 10.101 6/j. tine, A. M., Davis, A., & Barta, P. (2000). Elucidating the con- cogni tion. 2009. 05. 003 tributions of processing speed, executive ability, and frontal lobe Nasreddine, Z. S., Phillips, N. A., Bédirian, V., Charbonneau, S., volume to normal age-related differences in fluid intelligence. Whitehead, V., Collin, I., & Chertkow, H. (2005). The Montreal Journal of the International Neuropsychological Society, 6(1), Cognitive Assessment, MoCA: A brief screening tool for mild 52–61. https:// doi. org/ 10. 1017/ s1355 61770 06110 62 cognitive impairment. Journal of the American Geriatrics Soci- Segen, V., Avraamides, M. N., Slattery, T. J., & Wiener, J. M. (2020). ety, 53(4), 695–699. https:// doi. org/ 10. 1111/j. 1532- 5415. 2005. Age-related differences in visual encoding and response strategies 53221.x contribute to spatial memory deficits. Memory and Cognition. Nilakantan, A. S., Bridge, D. J., VanHaerents, S., & Voss, J. L. (2018). https:// doi. org/ 10. 3758/ s13421- 020- 01089-3 Distinguishing the precision of spatial recollection from its suc- Spaniol, J., Madden, D. J., & Voss, A. (2006). A diffusion model analy - cess: Evidence from healthy aging and unilateral mesial temporal sis of adult age differences in episodic and semantic long-term lobe resection. Neuropsychologia, 119, 101–106. https:// doi. org/ memory retrieval. Journal of Experimental Psychology: Learning 10. 1016/j. neuro psych ologia. 2018. 07. 035 Memory and Cognition, 32(1), 101–117. https://doi. or g/10. 1037/ Nuthmann, A. (2017). Fixation durations in scene viewing: Modeling 0278- 7393. 32.1. 101 the effects of local image features, oculomotor parameters, and Thapar, A., Ratcliff, R., & McKoon, G. (2003). A diffusion model task. Psychonomic Bulletin and Review, 24(2), 370–392. https:// analysis of the effects of aging on letter discrimination. Psychol- doi. org/ 10. 3758/ s13423- 016- 1124-4 ogy and Aging, 18(3), 415–429. https:// doi. org/ 10. 1037/ 0882- Olk, B., & Kingstone, A. (2015). Attention and ageing: Measuring 7974. 18.3. 415 effects of involuntary and voluntary orienting in isolation and Treitz, F. H., Heyder, K., & Daum, I. (2007). Differential course of in combination. British Journal of Psychology, 106(2), 235–252. executive control changes during normal aging. Aging, Neuropsy- https:// doi. org/ 10. 1111/ bjop. 12082 chology, and Cognition, 14(4), 370–393. https://doi. or g/10. 1080/ Pertzov, Y., Heider, M., Liang, Y., & Husain, M. (2015). Effects of 13825 58060 06784 42 healthy ageing on precision and binding of object location in vis- Verhaeghen, P., Cerella, J., & Basak, C. (2006). Aging, task com- plexity, and efficiency modes: the influence of working memory ual short-term memory. Psychology and Aging, 30(1), 26. https:// involvement on age differences in response times for verbal and doi. org/ 10. 1037/ a0038 396 1 3 420 Psychological Research (2022) 86:404–420 visuospatial tasks. Aging, Neuropsychology, and Cognition, 13(2), aging humans. Journal of Neuroscience, 33(14), 6012–6017. 254–280. https:// doi. org/ 10. 1080/ 13825 58909 69267https:// doi. org/ 10. 1523/ JNEUR OSCI. 0717- 12. 2013 Waldron-Perrine, B., & Axelrod, B. N. (2012). Determining an appro- Wilcox, R. R., & Keselman, H. J. (2003). Modem robust data analysis priate cutting score for indication of impairment on the Mon- methods: Measures of central tendency. Psychological Methods, treal Cognitive Assessment. International Journal of Geriatric 8(3), 254–274. https:// doi. org/ 10. 1037/ 1082- 989X.8. 3. 254 Psychiatry, 27(11), 1189–1194. https://doi. or g/10. 1002/ gps. 3768 Woods, D. L., Wyma, J. M., Yund, E. W., Herron, T. J., & Reed, B. Waller, D. (2006). Egocentric and nonegocentric coding in memory (2015). Age-related slowing of response selection and produc- for spatial layout: Evidence from scene recognition. Memory and tion in a visual choice reaction time task. Frontiers in Human Cognition, 34(3), 491–504. https://d oi.o rg/1 0.3 758/ BF0319 3573 Neuroscience, 9, 193. https://doi. or g/10. 3389/ fnhum. 2015. 00193 Waller, D. E., & Nadel, L. E. (2013). Handbook of spatial cognition. Zimmerman, S., Hasher, L., & Goldstein, D. (2011). Cognitive ageing: American Psychological Association. https:// doi. org/ 10. 1037/ a positive perspective. The paradoxical brain, 130–150 13936- 000 West, R. L. (1996). An application of prefrontal cortex function theory Publisher’s Note Springer Nature remains neutral with regard to to cognitive aging. Psychological Bulletin, 120(2), 272. https:// jurisdictional claims in published maps and institutional affiliations. doi. org/ 10. 1037/ 0033- 2909. 120.2. 272 Wiener, J. M., de Condappa, O., Harris, M. A., & Wolbers, T. (2013). Maladaptive bias for extrahippocampal navigation strategies in 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Psychological Research Springer Journals

Age-related changes in visual encoding strategy preferences during a spatial memory task

Loading next page...
 
/lp/springer-journals/age-related-changes-in-visual-encoding-strategy-preferences-during-a-mqfuX8lD7F

References (62)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2021
ISSN
0340-0727
eISSN
1430-2772
DOI
10.1007/s00426-021-01495-5
Publisher site
See Article on Publisher Site

Abstract

Ageing is associated with declines in spatial memory, however, the source of these deficits remains unclear. Here we used eye-tracking to investigate age-related differences in spatial encoding strategies and the cognitive processes underlying the age-related deficits in spatial memory tasks. To do so we asked young and older participants to encode the locations of objects in a virtual room shown as a picture on a computer screen. The availability and utility of room-based landmarks were manipulated by removing landmarks, presenting identical landmarks rendering them uninformative, or by presenting unique landmarks that could be used to encode object locations. In the test phase, participants viewed a second picture of the same room taken from the same (0°) or a different perspective (30°) and judged whether the objects occupied the same or different locations in the room. We found that the introduction of a perspective shift and swapping of objects between encoding and testing impaired performance in both age groups. Furthermore, our results revealed that although older adults performed the task as well as younger participants, they relied on different visual encoding strategies to solve the task. Specifically, gaze analysis revealed that older adults showed a greater preference towards a more categorical encoding strategy in which they formed relationships between objects and landmarks. Introduction 2020; Montefinese et al., 2015; Muffato et al., 2019; Segen et al., 2020). Building on these studies, and to gain a more Successful navigation and orientation depend on our abil- detailed understanding of the factors that contribute to the ity to recognise familiar places across different perspectives performance decline, we use eye-tracking to investigate (Waller & Nadel, 2013). In the lab, this ability is typically potential age-related differences in visual encoding strate- assessed with tasks in which participants first encode an gies. Specifically, we are interested in whether young and array of objects or environmental features from one per- older adults rely on the same or different environmental cues spective and are then asked to indicate whether the array during place recognition. has changed when presented from a different perspective. Recently, Muffato et al. (2019) and Hilton et al. (2020) Studies using such paradigms have reported age-related investigated the effects of cognitive ageing on place recog- declines in performance (Hartley et al., 2007; Hilton et al., nition abilities using scenes defined by objects that were placed in an open field. After encoding a scene with four objects, participants were presented with another scene * Vladislava Segen from a different perspective and had to decide whether or vladislava.segen@gmail.com not it was identical to the one encoded. Results revealed the Ageing and Dementia Research Centre, Bournemouth presence of object-location binding errors, particularly in University, Bournemouth, UK older adults. That is, compared to younger participants, older Department of Psychology, Bournemouth University, adults found it harder to detect that two objects had swapped Bournemouth, UK locations than when one of the objects was replaced with a Department of Psychology, University of Cyprus, Nicosia, new object. Cyprus In our previous work (Segen et al., 2020), we investi- CYENS Centre, Nicosia, Cyprus gated age-related differences in the ability to recognise spa- Aging and Cognition Research Group, German Center tial configurations across different perspectives. The task for Neurodegenerative Diseases (DZNE), Magdeburg, required participants to encode the locations of an array Germany Vol:.(1234567890) 1 3 Psychological Research (2022) 86:404–420 405 of identical objects presented as an image on a computer of landmarks. Second, the landmarks used in the environ- screen. The objects were arranged in clusters of one, two ment were all unique and informative and could, therefore, and three objects, in a virtual room containing additional facilitate the encoding of the object locations, even if they environmental cues such as windows and a door. Then, par- distracted the older participants. Third, there was substantial ticipants viewed a second image of the same room taken overlap between the landmarks and the objects of the scene, from the same (0°) or a different perspective (45° or 135°) which prevented a region of interest analysis. Finally, due and judged whether or not the objects were in the same to the large perspective shifts introduced in some trials (e.g. locations. The positions of the objects were either changed 135°), some landmarks were visible during encoding but by swapping two object clusters or by rotating one of the not at test. clusters. While with the former manipulation the task could The current study was designed to disentangle the expla- be solved using a coarse categorical representation of the nations for age-related differences in place recognition by spatial relationships between object clusters (e.g. the cluster examining gaze behaviour. To do so, we amended our origi- with two objects is to the left of the single object), the latter nal task (Segen et al., 2020) in a variety of ways to overcome manipulation required a fine-grained spatial representation the limitations of the earlier study. First, we reduced the size of the exact positions of the objects as the overall relation- of the perspective shift between encoding and test which ships between the clusters was maintained. allowed us to present the same landmarks during learning Consistent with previous research, we found that older and test, ensuring that participants could use the informa- adults had greater difficulty with the task than younger adults tion they encoded during learning to solve the task at test. (Hartley et al., 2007; Hilton et al., 2020; Montefinese et al., Decreasing the size of the perspective shift also made the 2015; Muffato et al., 2019). Diffusion modelling showed that task easier (Hegarty & Waller, 2004; Montofinese et al., older adults not only had greater difficulty in extracting use- 2015; Segen et al., 2020; Muffato et al., 2019; Hilton et al., ful information from the stimuli but that they also adopted a 2020). Task difficulty was further reduced by including only more conservative response strategy, i.e. they accumulated the condition in which two object clusters were swapped more information before reaching a decision. with each other. Reducing task difficulty aimed at avoiding Furthermore, the analysis of gaze data in Segen et al. floor level performance in older adults, which would allow (2020) revealed that older adults attended to a larger propor- us to rule out that potential differences in gaze behaviour tion of the scenes compared to younger adults. We proposed across groups are caused by participants’ inability to carry two potential explanations for this. First, differences in gaze out the task. behaviour may reflect differences in encoding strategies with Generally, we predict a decline in performance in older older adults encoding object locations relative to the land- adults consistent with age-related place recognition deficits marks available in the room (windows, door, etc.), whilst (Hartley et al., 2007). Responding after a perspective shift young adults focus on the local arrangement of objects and requires additional and demanding mental manipulations of on encoding the spatial relationships among them. The dif- the stored representations (e.g., mentally rotating the new ferences in encoding strategies may reflect a shift towards or the stored representation to match the other, imagining categorical spatial representations in older adults, driven moving around the array; Hegarty & Waller, 2004; Holmes by age-related hippocampal neurodegeneration (Antonova et al., 2018; King et al., 2002). Therefore, we expect that the et al., 2009; Meulenbroek et al., 2004; Moffat et al., 2007). introduction of the perspective shift would impair perfor- Second, older adults may have difficulties in focusing mance in both groups. However, we predict a larger decrease on the task-relevant information as they become distracted in performance in older adults who seem to have difficul- by salient features within the environment. This is in line ties with initiating those mental manipulations as reflected with the attention inhibition deficit in ageing reported in in past findings documenting larger impairments with the past studies (e.g., Hasher & Zacks, 1988). According to this introduction rather than the increase of the perspective shift account, older adults exhibit top-down control difficulties, (Montofinesse et al., 2015; Muffato et al., 2019; Hilton et al., with attention orienting being more affected by stimulus 2020; Segen et al., 2020). properties rather than the task at hand (Olk & Kingstone, To investigate the role of landmarks in encoding 2015; West, 1996). Lastly, older adults may have difficulties strategies and performance, we included trials in which in selecting the appropriate information required to solve the landmarks (in the form of posters on the walls) were: (1) task. This is consistent with our findings that older adults unique and could be used to encode object locations, (2) have difficulties in extracting useful information from the identical and thus uninformative or (3) absent from the stimuli (Segen et al., 2020). scene. Varying the availability and utility of room-based In our earlier study (Segen et al., 2020), we could not landmarks allowed us to test whether age-related differ - distinguish between these explanations for several reasons. ences in gaze behaviour during spatial encoding were due First, we did not systematically manipulate the availability to older adults encoding object positions by relating them 1 3 406 Psychological Research (2022) 86:404–420 to the landmarks or to older adults having difficulties in Virtual environment selecting and/or focusing on task-relevant information. Since this part of the study is largely exploratory, we The virtual environment was designed with Adobe 3DS have formulated a series of predictions about results that Max 2018 and depicted a 13.5  m × 14.6  m rectangular we would expect to find depending on how older adults room. The room contained 6 identical objects; pink vases use additional landmarks during encoding of object loca- on metal stands that were arranged in three clusters of 1, 2 tions. Given that the task can be solved either by focusing and 3 objects in the centre of the room (see Fig. 1). In the on the local arrangement of objects or by relating object No Landmarks condition, the walls contained no additional positions to landmarks, we should not necessarily expect cues, in the Uninformative Landmarks condition eight iden- age-related differences in performance if older adults sim- tical posters of the Tower Bridge were presented, two on ply shift towards a particular encoding strategy depending each wall. Finally, in the Informative Landmarks condition on which information is available. However, if older adults eight unique posters were presented, again two on each wall. select an encoding strategy that depends on the availability These posters consisted of highly familiar and recognisable of landmarks as suggested by our previous research (Segen landmarks (Hamburger & Röser, 2014): the Leaning Tower et al., 2020), we expect them to perform better when land- of Pisa, Stonehenge, the Statue of Liberty, the Golden Gate marks are informative than uninformative. Finally, if older Bridge, the Eiffel Tower, the White House, the Big Ben, and adults have difficulties focusing on task-relevant informa - the Great Wall of China. tion as a result of an attention inhibition deficit (Hasher & The experimental stimuli were renderings of the environ- Zacks, 1988), and are therefore distracted by the presence ment with a 70° horizontal field of view (FOV) with a 15% of landmarks, we predict worse performance when land- downward shift in the vertical FOV, yielding an asymmet- marks are available (either informative or uninformative) ric viewing frustum to simulate human vision. The virtual than when they are not. cameras from which the static images of the scenes were In terms of gaze behaviour, if older adults rely more on rendered were arranged on a circle (radius of 6.7 m) at 30° landmarks as part of their encoding strategy, compared to intervals, providing 12 possible camera positions and the their younger counterparts, we expect them to spend more object clusters were arranged in six unique layouts within time gazing at informative landmarks than uninformative the room (Fig. 1b-d). Six of those camera positions were landmarks. If, however, older adults are distracted by the used in the learning phase and in the 0° perspective shift landmarks, we expected them to show similar gaze behav- condition. The remaining 6 viewpoints were used in the iour in conditions with informative and uninformative test phase in the 30° perspective shift condition. Stimuli landmarks. were presented as static images on a desktop computer with OpenSesame 3.1.7 (Mathôt et al., 2012) and a standard com- puter keyboard was used to collect responses. Method Eye‑tracking Participants Eye movements were recorded using an Eyelink II (SR Research) head-mounted eye tracker at a rate of 500 Hz. Twenty-eight young (mean age = 21.00 years, SD = 2.27; age Calibrations were performed at least three times and drift range = 18–27 years; 15 females and 13 males) and 32 older correction was performed before each trial. The experiment adults aged 60 years and over (mean age = 68.80, SD = 6.34, was presented on a 102 cm screen (diagonal) with an aspect age range = 60–85; 17 females and 15 males) took part in ratio of 16:9 and a resolution of 1920 × 1080 pixels. Par- this study. Participants were recruited either through the ticipants were seated 100 cm from the monitor resulting in participant recruitment system of Bournemouth University a physical horizontal FOV of 47.9° and 28° vertical FOV. or through opportunity sampling in the community. Older adults received monetary compensation for their time whilst Procedure younger participants received course credits. Participants were screened for mild cognitive impairment using the Each experimental trial started with a fixation cross and a Montreal Cognitive Assessment (MoCA; Nasreddine et al., scrambled stimuli mask presented for 1500 ms (Fig. 1a). In 2005). Based on a threshold score of 23, no participants the learning phase, participants were presented with a ren- were excluded (Luis et al., 2009; Waldron-Perrine & Axel- dering of one of the 6 unique configurations of the target rod, 2012). All participants gave their written informed con- objects from one of the six possible viewpoints for 12 s. sent in accordance with the Declaration of Helsinki (World After this learning phase, participants were again presented Medical Association 2013). with a fixation cross and a scrambled stimuli mask for 1 3 Psychological Research (2022) 86:404–420 407 Fig. 1 a Experimental protocol; b, c and d Virtual environment and arrow) and swapping of the two object clusters (Black arrow) in stimuli for the experimental task, Blue and Green cameras repre- Swap trials (b, c). The middle and right panels show the two corre- sent the possible virtual cameras positions for the Learning and Test sponding snapshots for the learning and test phases, respectively. In b phase, respectively. Examples of possible object cluster layouts are and d there is a 30° perspective shift, to the left and right respectively. shown in b (No Landmarks), c (Uninformative Landmarks) and d In c there is no perspective shift. The black arrows in the right panel (Informative Landmarks). The left panel shows a survey perspective (b, c) indicate which clusters were swapped on the test stimuli of the example trials, indicating the rotation of the camera (Orange 1 3 408 Psychological Research (2022) 86:404–420 1500 ms. Then, in the test phase they were presented with a response time distribution. Prior to transforming, response rendering of the room either from the same viewpoint (50% times below 200 ms and over 20,000 ms were removed. of trials, Fig.  1c) or a different viewpoint that was offset by 30° from the study viewpoint (Fig. 1b, d). Participants were asked to respond by pressing the x or m keys on the Results keyboard as to whether the target objects were in the same locations as during the training phase or not. In 50% of the Accuracy trials, the target objects remained in the same locations (Same, Fig. 1c) and in the other 50% of the trials, two of the Accuracy estimates were obtained for each participant with three object clusters swapped locations (Swap, Fig. 1b, c). Age Group, Perspective Shift, Landmark Type and Manipu- As a result, chance level performance for this task was 50%. lation as fixed factors and a random by-subject and by-item The experiment consisted of 144 experimental trials that intercept. Coefficients, standard errors and z -values (Table 1) were preceded by 6 practice trials. The entire study took indicate that Perspective Shift and Manipulation affected around 90 min to complete and participants were allowed to performance. Specifically, accuracy decreased with the take breaks when they wished. introduction of a 30° Perspective Shift (Fig. 2a) and in the Swap condition (Fig. 2b). In addition, there was an inter- action between Perspective Shift and Manipulation with a Design greater decline in performance in the No Change condition compared to the Swap condition following a 30° Perspective The experiment followed a mixed 2 (Age Group: young vs. Shift (Fig.  2c). Finally, we found a three-way interaction older adults) × 2 (Manipulation: Same, Swap, Fig. 1b,c and between Perspective Shift, Manipulation and Age Group d) × 2 (Perspective Shift: 0°, 30°) × 3 (Landmark Type: No with older adults showing a larger decline in performance Landmarks, Uninformative, Informative) design with Manip- than younger adults in the No Change condition when a ulation, Perspective Shift and Landmark Type manipulated 30° Perspective Shift was introduced, whilst displaying an within participants and Age Group between. increase in performance in the Swap condition when a 30° Perspective Shift was introduced (Fig. 2d). Effect plots for Data Analysis significant main effects and interactions are reported in the Supplementary Materials. Data from one older participant were excluded from all analyses due to chance level performance in the 0° Perspec- Response Time tive Shift condition. The remaining data were analysed with linear mixed-effects models (LME) using LME4 (Bates As with accuracy, response time estimates were obtained for et al., 2018) in R (R Core Team, 2013). Specifically, accu- each participant with Age Group, Perspective Shift, Landmark racy was analysed using generalized linear mixed-effects Type and Manipulation as fixed factors and a random by-sub- (GLME) models with the glmer function from LME4 pack- ject and by-item intercept with a random slope for Manipu- age. The following contrasts were used in all (G)LMEs con- lation across participants. Coefficients, standard errors and ducted: Age Group (Younger adults/Older adults), Perspec- t-values (Table 2) show that Age Group, Perspective Shift, tive shift (0°/30°) and Manipulation (No Change/Swap) were Landmark Type and Manipulation were all reliable predic- coded using effect coding. This coding scheme compares tors of response time. Specifically, we found that older adults the effect of a variable (i.e. Age Group) on performance were slower to respond compared to younger adults (Fig. 3a), averaged across all levels of other variables (i.e. Perspec- and that response times increased with the introduction of tive Shift and Manipulation). Landmark Type was coded a Perspective Shift (Fig.  3b). In addition, response times using treatment coding. Since we were interested in examin- were longer with Informative than Uninformative Landmark ing the difference between Informative and Uninformative Type (Fig. 3c) and in the Swap condition compared to the No Landmarks and the difference between No Landmarks and Change condition (Fig. 3d). We also found a significant inter - Uninformative Landmarks, we used the Uninfomative Land- action between Age Group and Manipulation with a smaller mark as the baseline. As a result, all of the effects for other increase in response times in the Swap condition in older than factors are calculated with reference to the performance in younger adults (Fig. 3e). There was also a Perspective Shift and the Uninformative Landmark, rather than the average of per- Manipulation interaction with a smaller increase in response formance for all levels of Landmark Type. For the response times in the Swap condition than the No Change condition time analysis, we included only the correct trials and we log- with the introduction of the Perspective Shift (Fig. 3f). We also transformed response times following the recommendations found an interaction between Landmark Type and Manipula- of Baayen et al. (2008) for dealing with the skewness of the tion with a smaller increase in response times between the No 1 3 Psychological Research (2022) 86:404–420 409 Table 1 Coefficients from Accuracy GLME analysis Predictors Accuracy Coefficients Std. Error z-value (Intercept) 2.023 0.262 7.724 Age Group (Old) 0.145 0.112 1.293 Perspective Shift (30°) − 0.635 0.079 − 8.049 Landmark Type (Informative) − 0.122 0.347 − 0.350 Landmark Type (No Landmarks) 0.066 0.350 0.189 Manipulation (Swap) − 1.316 0.086 − 15.216 Age Group (Old): Perspective Shift (30°) − 0.104 0.071 − 1.468 Age Group (Old): Landmark Type (Informative) − 0.063 0.095 − 0.659 Age Group (Old): Landmark Type (No Landmarks) − 0.138 0.105 − 1.314 Perspective Shift (30°): Landmark Type (Informative) 0.176 0.106 1.659 Perspective Shift (30°): Landmark Type (No Landmarks) − 0.037 0.116 − 0.319 Age Group (Old): Manipulation (Swap) 0.063 0.071 0.887 Perspective Shift (30°): Manipulation (Swap) 0.414 0.077 5.387 Landmark Type (Informative): Manipulation (Swap) 0.212 0.115 1.846 Landmark Type (No Landmarks): Manipulation (Swap) − 0.082 0.125 − 0.656 Age Group (Old): Perspective Shift (30°): Landmark Type (Informative) 0.097 0.095 1.020 Age Group (Old): Perspective Shift (30°): Landmark Type (No Landmarks) 0.137 0.105 1.303 Age Group (Old): Perspective Shift (30°): Manipulation (Swap) 0.240 0.071 3.399 Age Group (Old): Landmark Type (Informative): Manipulation (Swap) 0.049 0.095 0.514 Age Group (Old): Landmark Type (No Landmarks): Manipulation (Swap) 0.054 0.105 0.512 Perspective Shift (30°): Landmark Type (Informative): Manipulation (Swap) 0.060 0.103 0.584 Perspective Shift (30°): Landmark Type (No Landmarks): Manipulation (Swap) 0.155 0.114 1.364 Age Group (Old): Perspective Shift (30°): Landmark Type (Informative): Manipulation (Swap) − 0.122 0.095 − 1.277 Age Group (Old): Perspective Shift (30°): Landmark Type (No Landmarks): Manipulation (Swap) − 0.201 0.105 − 1.916 Significant z values (|z|≥ 1.96) in bold Fig. 2 Bar plots of accuracy values for a significant main effect of a Perspective Shift, b Manipulation, and sig- nificant interactions c between Manipulation and Perspective Shift and d Interaction between Age Group, Manipulation and Perspective Shift with a mean (solid line) and 95% CIs (grey shaded area) with individual data points and violin plots behind 1 3 410 Psychological Research (2022) 86:404–420 Table 2 Coefficients from response time LME analysis Predictors Log transformed response time Estimates Std. Error t-value (Intercept) 7.834 0.041 190.067 Age Group 0.209 0.040 5.248 Perspective Shift (30°) 0.130 0.015 8.459 Landmark Type (Informative) 0.058 0.020 2.942 Landmark Type (No Landmarks) − 0.013 0.020 − 0.640 Manipulation (Swap) 0.133 0.011 12.386 Age Group: Perspective Shift (30°) 0.006 0.014 0.451 Age Group: Landmark Type (Informative) 0.019 0.013 1.470 Age Group: Landmark Type (No Landmarks) − 0.019 0.013 − 1.443 Perspective Shift (30°): Landmark Type (Informative) 0.012 0.015 0.813 Perspective Shift (30°): Landmark Type (No Landmarks) − 0.000 0.015 − 0.007 Age Group: Manipulation (Swap) − 0.032 0.009 − 3.474 Perspective Shift (30°): Manipulation (Swap) − 0.077 0.010 − 7.542 Landmark Type (Informative): Manipulation (Swap) − 0.034 0.015 − 2.259 Landmark Type (No Landmarks): Manipulation (Swap) 0.010 0.015 0.654 Age Group: Perspective Shift (30): Landmark Type (Informative) − 0.003 0.013 − 0.239 Age Group: Perspective Shift (30°): Landmark Type (No Landmarks) − 0.013 0.013 − 1.012 Age Group: Perspective Shift (30°): Manipulation (Swap) − 0.018 0.009 − 1.960 Age Group: Landmark Type (Informative): Manipulation (Swap) − 0.008 0.013 − 0.596 Age Age Group: Landmark Type (No Landmarks): Manipulation (Swap) − 0.024 0.013 − 1.847 Perspective Shift (30°): Landmark Type (Informative): Manipulation (Swap) 0.019 0.014 1.312 Perspective Shift (30°): Landmark Type (No Landmarks): Manipulation (Swap) 0.002 0.014 0.162 Age Group: Perspective Shift (30°): Landmark Type (Informative): Manipulation (Swap) 0.005 0.013 0.406 Age Group: Perspective Shift (30°): Landmark Type (No Landmarks): Manipulation (Swap) − 0.004 0.013 − 0.289 Significant t values (|t|≥ 1.96) in bold Change and the Swap condition in the Informative Landmark Theory evaluates sensitivity and response bias in situa- Type (Fig. 3g) compared to Uninformative Landmark Type tions that require decision making under uncertainty. It condition. Finally, we found a three-way interaction between is applied when a binary decision about the presence or Age Group, Perspective Shift and Manipulation, with the Age absence of a signal is made, comparing the response with Group and Perspective Shift interactions showing a different the actual presence/absence of the signal. With Signal trend across No Change and Swap Manipulation. Specifically, Detection Theory, the formula c = -0.5[z(hit rate) + z(false there was a larger increase in response times in older adults alarm rate) is used to compute response bias, where hit than young adults, in the No Change condition with the intro- rate and false alarm rates refer to trials in which the sig- duction of the Perspective Shift (Fig. 3). Whilst in the Swap nal was correctly or incorrectly, respectively, reported as condition, the increase in response times in older adults was present. smaller when a Perspective Shift was introduced compared Overall, there was a positive response bias showing to young adults. Effect plots for significant main effects and that participants were more likely to respond that nothing interactions are reported in the Supplementary Materials. has changed than to respond that something had changed (Fig.  4). LMM analysis (Table  3) with Age Group, Per- spective Shift and Landmark Type as fixed factors and by- Response Bias subject intercept with a random slope for Perspective Shift, indicated that the introduction of a Perspective Shift led To examine if participants displayed a response bias, we to a decrease in response bias, which was larger in older carried out an analysis based on Signal Detection Theory adults than in younger adults. Furthermore, when a Per- (Harvey, 1992; Macmillan & Creelman, 1991) using the spective Shift was introduced, the response bias decreased sdt.rmcs (Todorova, 2017) package in R. Signal Detection more in the No Landmarks and the Informative Landmarks 1 3 Psychological Research (2022) 86:404–420 411 Fig. 3 Bar plots of Response Times values for significant main effects only for Landmark Type (Informative): Manipulation (Swap)) H: Age of A: Age Group B: Perspective Shift C: Landmark Type (signifi- Group, Manipulation and Perspective Shift with mean (solid line) and cant only for the Informative Landmark Type) D: Manipulation and 95% CIs (grey shaded area) with individual data points and violin interactions between E: Age Group and Manipulation F: Perspective plots behind Shift and Manipulation G: Landmark Type and Condition (significant 1 3 412 Psychological Research (2022) 86:404–420 Fig. 4 Bar plots for Response Bias as a function of Age Group, Landmark Type and Per- spective Shift with mean (solid line) and 95% CIs (grey shaded area) with individual data points and violin plots behind. Stars indicate response bias signifi- cantly different from 0 (1 star [p < 0.05], 2 stars [p < 0.01] and 3 stars [p < 0.001]) Table 3 Coefficients from Response Bias LME analysis Predictors Response bias (c) Estimates Std. Error t-value (Intercept) 0.437 0.033 13.043 Age Group (Older Adults) − 0.047 0.033 − 1.403 Perspective Shift (30°) − 0.069 0.029 − 2.384 Landmark Type (Informative) 0.003 0.026 0.097 Landmark Type (No Landmarks) − 0.048 0.026 − 1.826 Age Group: Perspective Shift (30°) − 0.072 0.029 − 2.495 Age Group: Landmark Type (Informative) 0.007 0.026 0.264 Age Group: Landmark Type (No Landmarks) − 0.008 0.026 − 0.306 Perspective Shift (30°): Landmark Type (Informative) − 0.052 0.026 − 1.978 Perspective Shift (30°): Landmark Type (No Landmarks) − 0.058 0.026 − 2.201 Age Group: Perspective Shift (30°): Landmark Type (Informative) 0.049 0.026 1.845 Age Group: Perspective Shift (30°): Landmark Type (No Landmarks) 0.028 0.026 1.043 Significant t values (|t|≥ 1.96) in bold conditions compared to the Uninformative Landmarks con- Gaze analysis dition (Table 4). Fixations and saccades were identified using the SR Research algorithms and were pre-processed as follows: First, we removed fixations that contained a blink, fell 1 3 Psychological Research (2022) 86:404–420 413 Table 4 Coefficients from Dwell Time on the top IA LME analysis predictors of Dwell Time on the top IA. Specifically, we found that compared to the Uninformative Landmarks Predictors Dwell Time on Landmarks condition that was used as a baseline, there was a reduc- Estimates Std. Error t-value tion in Dwell Time on the top IA in the No Landmarks and an increase in Dwell Time in the Informative Landmarks (Intercept) 13.054 1.503 8.684 condition. We also found that older adults spent more time Age Group (Older Adults) 2.99/ 1.457 2.058 looking at the top IA compared to younger adults. In addi- Landmark Type (No Land- − 8.108 0.644 − 12.600 marks) tion, there was a Landmark Type and Age Group interaction Landmark Type (Informative) 9.540 0.644 14.826 whereby older adults’ Dwell Time on Landmarks decreased Age Group (Older Adults): − 1.804 0.375 − 4.812 more than that of younger adults’ in the No Landmarks con- Landmark Type (No Land- dition compared to Uninformative Landmarks condition and marks) showed a larger increase in the Informative Landmarks con- Age Group (Older Adults): 1.171 0.375 3.124 dition compared to the Uninformative Landmarks condition. Landmark Type (Informa- A Dwell Time analysis on the top IA at test produced similar tive) results to those of the learning phase, with the exception Significant t values (|t|≥ 1.96) in bold that the increase in Dwell Time in older adults and the Age Group by Landmark Type (No Landmarks) interaction were not significant. Results from this analysis are presented in outside of the screen boundaries or were shorter than 80 ms the Supplementary Materials. or longer than 1000 ms (Inhoff & Radach, 1998; Nuthmann, 2017). Finally, we removed saccades with maximum ampli- Relationship between Gaze and Performance tudes (41.35°va) or velocities (1500°/s) larger than it should be possible based on the distance of the participant from the Dwell time on the top IA was not related to performance screen and the screen size. across any of the three Landmark Type conditions (Fig. 6), The primary aim of the gaze analysis was to investigate suggesting that the task could be solved either by using age differences in encoding strategies and was therefore Landmarks (when they are available) or by focusing primar- mainly focused on the analysis of gaze during the encoding ily on the objects. Thus, the differences in gaze behaviour phase. Analysis of differences in basic saccade and fixation reported here are likely to represent differences in encoding parameters between young and older adults showed that strategy preferences that change with age. during the 12 s encoding period, older adults made shorter and more frequent fixations as well as more frequent sac- Gaze behaviour across trials cades. The results are reported in detail in the supplementary materials. We also investigated if gaze behaviour changes across time by correlating Dwell Time on landmarks with trial older Gaze on landmarks for younger and older participants in the No Landmark, Uninformative and Informative Landmark conditions. We As we were primarily interested in age-related differences found that across both younger and older adults, Dwell Time in gaze as a function of Landmark Type, we split stimuli remained consistent in the No Landmark condition through- into two interest areas (See Fig. 5) and compared the per- out the experiment (Young: r = 0.011, p = 0.895, Older: centage of Dwell Time on the top interest area (IA) where r = − 0.09, p = 0.279). In the Uninformative Landmark con- Landmarks were located when available vs. the bottom area dition, older adults spent less time fixating on landmarks where the objects were located. To do so, we computed the over the course of the experiment (r = − 0.18, p = 0.032), total dwell time for each trial by adding up the duration of whilst younger adults’ gaze (r = − 0.05, p = 0.543) remained all fixations in the trial. Next, we calculated the proportion unchanged. In the Informative Landmark condition, an oppo- of dwell time that was spent fixating in the top IA. This site pattern of results was found with younger adults spend- approach allowed us to specifically focus on age-related ing less time fixating on landmarks (r = − 0.20, p = 0.018) differences in the use of room-based Landmarks during with older adults’ gaze remaining unchanged (r = − 0.09, encoding with the increased Dwell Time on the upper part p = 0.266). of the stimuli largely reflecting gaze on Landmarks (when available). Consistency in gaze between learning and test LME analysis with Age Group and Landmark Type as fixed factors and a by-subject and by-item random intercept Finally, we examined if participants showed similar gaze showed that Landmark Type and Age Group were reliable behaviour during learning and test. To do so, we correlated the 1 3 414 Psychological Research (2022) 86:404–420 Fig. 5 Heatmaps representing number of fixations as a func- tion of age group and landmark type Dwell Time on the top IA across different Landmark Types at Discussion learning and test. We found strong positive correlations across all Landmark Types (No Landmarks: R2 = 0.67, p < 0.001; In the present study, we used eye-tracking to investi- Uninformative: R2 = 0.88, p < 0.001; Informative: R2 = 0.94, gate age-related differences in visual encoding strategies p < 0.001). Those correlations suggest that participants are employed for memorizing the locations of objects in a highly consistent in which stimulus features they gaze at dur- room. To do so, we explored if participants were able to ing encoding and test. 1 3 Psychological Research (2022) 86:404–420 415 of previous studies (Muffato et al., 2019; Montofinese et al., 2015; Segen et al., 2020). In addition, the scene at test could differ from the encoded only in terms of a change in the categorical relationship between objects. That is, in contrast to Segen et al. (2020), in the current study no changes in fine-grained spatial relationships between objects occurred. That the easier task may be responsible for the lack of age- related deficits in task accuracy is in line with cognitive age- ing research reporting greater age-related differences in per - formance with increasing task difficulty (Angel et al., 2016; Earles et al., 2004; Verhaeghen et al., 2006). The lack of age-related performance accuracy differences in less demanding tasks can be explained by the compensa- tion-related utilization of neural circuits hypothesis (Reuter- Lorenz & Cappell, 2008). This hypothesis posits that under low task demands older adults can perform the tasks as well as young adults, supported by increased neural activations. Fig. 6 Scatter Plot between Dwell Time on the top IA and Accuracy However, when task demands increase, older adults’ cogni- as a function of Landmark Type with regression line and CI (shaded tive limits are reached resulting in performance declines that area) are typically accompanied by a reduction in activation in the relevant neural networks (Morcom & Rugg, 2007; Angel identify whether a spatial scene has changed following a et al., 2016). Thus, it is plausible that due to the relatively perspective shift between encoding and test. The 30° per- low task-demands in the current study, which are reflected spective shift was introduced to ensure that participants in high performance across both age groups, older adults relied on spatial representations instead of solving the were able to carry out the task just as accurately as younger task by matching the visual image with a stored visual participants. snapshot from encoding (Nardini et al., 2009). To inves- Consistent with our predictions, we found declines in tigate the effect of landmarks on encoding strategies, we accuracy in both age groups that were accompanied by an also manipulated the availability and informative value of increase in response times when a perspective shift was landmarks within the environment. introduced. This reduction in performance may have been We found that overall, older adults took longer to respond. driven by qualitative differences between trials that involved This increase in response times is consistent with findings a perspective shift and those that did not. Specifically, with- that are widely reported in the cognitive ageing literature out a perspective shift participant can refer to the represen- (Choice reaction time task: Woods et al., 2015; Memory: tation of the learned scene from memory and use image Hertzog et al., 2003; Language: Ratcliff et al., 2004a, b), and matching to detect changes (Nardini et al., 2009). However, istypically attributed to decrements in speed of processing the introduction of the perspective shift required partici- (Salthous, 1996; Salthouse & Ferrer-Caja, 2003). We also pants to engage in additional cognitive processing related found that the introduction of the perspective shift and the to mental transformation in order to match the perspectives manipulation of object positions led to performance dec- of the stored spatial configuration with the one presented at rements in both age-groups. The availability and informa- test (Hegarty & Waller, 2004). However, it should be noted tiveness of the room-based landmarks did not affect task that the effect of the perspective shift was small, which is accuracy. Importantly, we found that when landmarks were likely due to the relatively small perspective shift that we presented, older participants spent more time than younger introduced. participants looking at the upper part of the display that con- Interestingly, there was a much more nuanced (if any) tained the landmarks. This was particularly the case when decline in accuracy or increase in response time in the Swap the landmarks were informative. compared to the No Change condition when a perspective Contrary to our expectations and previous place recogni- shift was introduced. To explain such findings, we turn to the tion research (Muffato et al., 2019; Hilton et al., 2020; Segen response bias analysis which suggested that the introduction et al., 2020; Harley et al., 2007), there were no age-related of the perspective shift increased the likelihood of partici- differences in accuracy. However, it should be noted that pants responding that the object positions were “different”. we used an easier task than those used in previous studies, Thus, when a perspective shift was introduced in the Swap which could yield fewer problems for older adults. For exam- condition, this led to an increase in the number of correct ple, the perspective shift we introduced was smaller than that responses albeit for the wrong reason. We believe that the 1 3 416 Psychological Research (2022) 86:404–420 increase of “different” responses after a perspective shift to use eye-tracking to investigate age-related differences arises from the salient change in the visual input indicating in spatial encoding strategies and to study if such differ - that “something is different”. However, if participants were ences are driven by the information available within the solely responding to any change in the visual information environment. Firstly, we focused on general gaze param- between encoding and test, we expected them to perform eters and found that older adults made more fixations that below chance level in the No Change condition when a per- were shorter in duration as well as shorter saccades than spective shift was present. Yet, our participants were still young adults. While these results are consistent with those able to perform well in this condition and their performance from a recent study using a similar place recognition task in the Swap condition with perspective shifts was not at the (Hilton et al., 2020), relating these general gaze measures ceiling. This pattern of results demonstrates that participants to encoding strategies is difficult. We thus performed IA were not solely relying on basic visual change detection but analysis which showed that gaze behaviour differed as a were instead using a spatial strategy to perform the task. function of room type. As expected, we found that both Yet, they might have found it hard to inhibit the immediate age groups spent the lowest amount of time looking at the response that the image is “not the same” when the perspec- upper part of the stimuli in the No Landmarks condition tive shift was introduced. The increase in performance in in which there were no images on the walls of the room, older adults with the introduction of the perspective shift in followed by the Uninformative Landmarks condition, in the Swap condition may thus be due to older adults expe- which the images on the walls were all identical, and the riencing even greater difficulty in inhibiting the response Informative Landmarks condition in which each image was that the image is “not the same” when a perspective shift unique. These findings are consistent with results reported was present. Such difficulties are in line with age-related by Livingstone-Lee et al. (2011) who showed that partici- decline in executive functioning, in particular executive pants quickly learned to adapt their gaze distribution in a control (Braver & West, 2008; Schretlen et al., 2000; Treitz virtual Morris water maze task based on the information et al., 2007). that was available in the environment. Importantly, we Overall participants were more likely to make errors in found that compared to younger adults, older adults spent the Swap condition than the No Change condition. To per- more time looking at landmarks in the Uninformative and form the task accurately participants in either condition had Informative Landmarks conditions during encoding. A to bind an object’s identity to its location (Postma et al., similar pattern was observed during the test phase in the 2004; Waller, 2006). Previous research has shown that this is Informative Landmarks condition. a cognitively demanding and error-prone process. For exam- A possible explanation for these age-related-differences ple, in place recognition studies participants were shown in gaze behaviour is that older adults simply look around to be less accurate in detecting that a change has occurred more due to a lack of a systematic encoding strategy. This when two objects swapped places compared to when a previ- can arise as a result of difficulties in selecting task-relevant ously shown object was replaced by a new one (Hilton et al., information (Raptis et al., 2017). Given our results, how- 2020; Muffato et al., 2019). Similar results are reported in ever, it appears unlikely that older adults were randomly visuospatial working memory studies in which participants scanning the environment without a clear encoding strat- were asked to encode positions of abstract objects on a blank egy for several reasons: first, older adults solved the task display. Participants were more likely to make swap errors, as accurately as younger participants, which would not be that is to place objects in the positions that were previously possible without a clear encoding strategy. Second, we found occupied by a different object (Pertzov et al., 2012, 2015). that older adults’ gaze behaviour changed as a function of Thus, the lower performance in the Swap condition can the landmarks used. Specifically, older adults spent signifi- be explained by difficulties with binding objects to their cantly more time looking at the upper part of the stimuli locations, which prevents participants from accumulating when landmarks were present and when these landmarks information signalling that a change has occurred (Hilton were informative, i.e. when they could be used to encode et al., 2020; Muffato et al., 2019). Specifically, in the current the spatial locations of the objects by relating objects to task, the objects within the scene and their general configu- these room-based landmarks. Third, both younger and older ration remained the same between learning and test. The adults adapted their gaze behaviour over the course of the only change introduced in the Swap condition is the posi- experiment such that older adults spent less time fixating on tion that each cluster occupied within that general configura- uninformative landmarks across trials. Younger participants, tion. Therefore, participants needed to remember the specific on the other hand, spent less time fixating on informative locations of each object cluster within that configuration to landmarks across the trial. These changes in gaze behaviour detect that a change has occurred. over time are likely to reflect adaptations of encoding strate- In addition to comparing the behavioural performance gies with older adults learning to inhibit attending to unin- of older and younger adults, another aim of this study was formative information and younger participants focusing on 1 3 Psychological Research (2022) 86:404–420 417 encoding the relationship between objects even in the pres- enough resources to deal with the task at hand and if they are ence of informative landmarks. directing already limited resources to task-irrelevant infor- Finally, gaze behaviour was highly consistent between mation (Angel et al., 2016; Morcom et al., 2007; Reuter- learning and test, which suggests that participants, both Lorenz & Cappell, 2008). young and older, attended to the same information during The idea that older adults have a greater preference than learning and test. It is possible that low-level properties of young adults towards encoding strategies that incorporate all the stimuli (i.e. colour, intensity and orientation) contributed available landmarks is consistent with results from research to such similarities in gaze behaviour through bottom-up that employs diffusion modelling. Several studies document control of attention (Itti, 2005), as similar visual information an age-related shift towards a more conservative response was presented at both learning and test. However, given that strategy whereby, compared to young adults, older adults participants performed well on the task and made very few prefer to accumulate more information before making deci- fixations at test, it is unlikely that the consistency between sions (Ratcliff et al., 2006, 2004a, b; Segen et al., 2020; gaze behaviour at learning and test was solely driven by Spaniol et  al., 2006; Thapar et  al., 2003). This explana- bottom-up processes. Instead, we suggest that participants tion is also supported by our findings of longer response relied on the information they encoded at learning to make times in older adults which could be indicative of greater decisions regarding whether or not the objects have moved cautiousness. at test. Together, these results suggest that gaze behaviour, Alternatively, the preference for attending to landmarks in both younger and older adults, represents task and stim- during encoding could be indicative of age-related differ - uli-dependent visual strategies that participants employed ences in spatial encoding strategies. Specifically, older to solve the task. adults’ may be more reliant on an encoding strategy in Age-related differences in gaze behaviour may also be which they relate the positions of objects to landmarks, driven by older adults being distracted by salient, but task- while younger participants focus on the local arrangement irrelevant landmarks as a result of attention inhibition defi- of objects and encode the spatial relationships between cits (Hasher & Zack, 1988; Healey et al., 2008, 2013). This them. This interpretation is in line with our findings that account is partly supported by our findings as older adults older adults spent more time than younger adults looking spent more time than younger adults gazing at the uninform- at the landmarks during encoding, especially when these ative landmarks. Notably, however, this did not affect their were informative. The differences in encoding strategies may performance and can be explained by the relatively long represent an age-related shift towards the use of a categori- encoding times that allowed participants to encode adequate cal encoding strategy whereby participants bind an object task-relevant information even if they were briefly distracted. to the nearest cue/landmark without the need to encode the An alternative explanation as to why older adults attended exact metric relationship between the two. This shift may to uninformative landmarks (i.e. task-irrelevant informa- arise from difficulties in forming precise spatial representa- tion), is a more general age-related shift in the way they tions. For example, previous visuospatial working memory approach cognitive tasks. Zimmerman et al. (2011) sug- research has shown that older adults were less precise in esti- gested that older adults tend to implicitly encode all of the mating previous locations of objects compared to younger available information, regardless of its immediate utility. adults, despite positioning the objects in the correct region This is consistent with evidence showing that the inability to of the stimuli (Nilakantan et al., 2018; Pertzov et al., 2015). inhibit attention sometimes comes with benefits. Kim et al. Furthermore, in navigation, older adults show greater pref- (2007), for example, have shown that older adults display erence towards the use of beacon strategies (Wiener et al., greater priming benefits when distractors on a previous task 2013). Such strategies involve coarse categorical represen- were used as primes in a problem-solving task. It is pos- tations of locations in relation to environmental beacons or sible that the shift towards encoding irrelevant, as well as landmarks and may be preferred by older adults due to dif- relevant information, stems from greater experience with ficulties in formulating more precise representations. real-world environments in which apparently task-irrelevant Lastly, we did not find any relation between gaze behav - information often becomes relevant in the future (Kim et al., iour and performance. This is not surprising as we found 2007; Zimmerman et al., 2011). For example, remembering similar performance across different room types and across extra landmarks in the environment could help to distinguish both age groups despite the presence of gaze differences. similar environments from each other. Such implicit shifts in These results indicate that the current task can be solved encoding strategies may explain why older adults spent more equally well by focusing on objects and by relating the time looking at extra information even if this information objects to landmarks (if they are available), with older adults is not strictly necessary for solving the task at hand. How- showing a preference towards the latter. In addition, the lack ever, such strategy shifts could lead to performance deficits of correlation between gaze and performance is consistent in cognitively taxing situations, if older adults do not have with our previous findings showing that the Swap condition 1 3 418 Psychological Research (2022) 86:404–420 Neural correlates of successful memory retrieval in aging: do could be solved either by looking around more or by having executive functioning and task difficulty matter? Brain Research, more focused gaze (Segen et al., 2020) outlining that coarse 1631, 53–71. https:// doi. org/ 10. 1016/j. brain res. 2015. 10. 009 spatial representations can be formed using a wider range of Antonova, E., Parslow, D., Brammer, M., Dawson, G. R., Jackson, S. encoding strategies and the available information. H. D., & Morris, R. G. (2009). Age-related neural activity during allocentric spatial memory. Memory, 17(2), 125–143. https://doi. To summarise, our results suggest that under specific con- org/ 10. 1080/ 09658 21080 20773 48 ditions such as the presence of a relatively small perspective Baayen, R. H., Davidson, D. J., & Bates, D. M. (2008). Mixed-effects shift and the introduction of categorical changes within the modeling with crossed random effects for subjects and items. scene, spatial memory is resistant to age-related changes Journal of Memory and Language, 59(4), 390–412. https:// doi. org/ 10. 1016/j. jml. 2007. 12. 005 as older adults perform the task as well as younger partici- Braver, T. S., & West, R. (2008). Working memory, executive con- pants. Furthermore, we report an age-related shift in visual trol, and aging. In F. I. M. Craik and T. A. Salthouse (Eds.), The encoding strategy. Although we cannot completely rule out handbook of aging and cognition (p. 311–372). Psychology Press that these changes in gaze behaviour are driven by inhibi- Earles, J. L., Kersten, A. W., Berlin Mas, B., & Miccio, D. M. (2004). Aging and memory for self-performed tasks: effects of task dif- tory control mechanisms, it seems highly plausible that older ficulty and time pressure. The Journals of Gerontology Series adults, who might be more distracted by the uninformative B: Psychological Sciences and Social Sciences, 59(6), 285–293. landmarks, employ an encoding strategy that relies on pro- https:// doi. org/ 10. 1093/ geronb/ 59.6. P285 cessing the categorical relationships between objects and Hamburger, K., & Röser, F. (2014). The role of landmark modality and familiarity in human wayfinding. Swiss Journal of Psychology. room-based landmarks rather than forming fine-grain spatial https:// doi. org/ 10. 1024/ 1421- 0185/ a0001 39 representations. Hartley, T., Bird, C. M., Chan, D., Cipolotti, L., Husain, M., Varga- Khadem, F., & Burgess, N. (2007). The hippocampus is required Supplementary Information The online version contains supplemen- for short-term topographical memory in humans. Hippocampus, tary material available at https://doi. or g/10. 1007/ s00426- 021- 01495-5 . 17(1), 34–48. https:// doi. org/ 10. 1002/ hipo. 20240 Hasher, L., & Zacks, R. T. (1988). Working Memory, Comprehension, and Aging: A Review and a New View. Psychology of Learning Compliance with ethical standards and Motivation - Advances in Research and Theory, 22, 193–225. https:// doi. org/ 10. 1016/ S0079- 7421(08) 60041-9 Conflict of interest The authors declare that the research was con- Healey, M. K., Campbell, K. L., & Hasher, L. (2008). Cognitive aging ducted in the absence of any commercial or financial relationships that and increased distractibility: Costs and potential benefits. Progress could be construed as a potential conflict of interest. in Brain Research, 169, 353–363. https://doi. or g/10. 1016/ S0079- 6123(07) 00022-2 Ethical standards This study was carried out in accordance with the Healey, M. K., Hasher, L., & Campbell, K. L. (2013). The role of recommendations of the Research Ethics Code of Practice, Science, suppression in resolving interference: evidence for an age-related Technology and Health Research Ethics Panel at Bournemouth Univer- deficit. Psychology and Aging, 28(3), 721–728. https:// doi. org/ sity with written informed consent from all subjects and has therefore 10. 1037/ a0033 003 been performed in accordance with the ethical standards laid down in Hegarty, M., & Waller, D. (2004). A dissociation between mental rota- the 1964 Declaration of Helsinki and its later amendments. tion and perspective-taking spatial abilities. Intelligence, 32(2), 175–191. https:// doi. org/ 10. 1016/j. intell. 2003. 12. 001 Open practices statement The datasets generated during and/or Hertzog, C., Dixon, R. A., Hultsch, D. F., & MacDonald, S. W. (2003). analysed during the current study are available in the Open Science Latent change models of adult cognition: are changes in process- Framework repository, https://osf .io/ v4m we/. This experiment was not ing speed and working memory associated with changes in epi- pre-registered. sodic memory? Psychology Aging, 18(4), 755. https://doi. or g/10. 1037/ 0882- 7974. 18.4. 755 Hilton, C., Muffato, V., Slattery, T. J., Miellet, S., & Wiener, J. (2020). Open Access This article is licensed under a Creative Commons Attri- Differences in encoding strategy as a potential explanation for bution 4.0 International License, which permits use, sharing, adapta- age-related decline in place recognition ability. Frontiers in Psy- tion, distribution and reproduction in any medium or format, as long chology, 11, 2182. https:// doi. org/ 10. 3389/ fpsyg. 2020. 02182 as you give appropriate credit to the original author(s) and the source, Holmes, C. A., Newcombe, N. S., & Shipley, T. F. (2018). Move to provide a link to the Creative Commons licence, and indicate if changes learn: Integrating spatial information from multiple viewpoints. were made. The images or other third party material in this article are Cognition, 178, 7–25. https:// doi. org/ 10. 1016/j. cogni tion. 2018. included in the article’s Creative Commons licence, unless indicated 05. 003 otherwise in a credit line to the material. If material is not included in Inhoff, A. W., & Radach, R. (1998). Definition and computation of the article’s Creative Commons licence and your intended use is not oculomotor measures in the study of cognitive processes. In G. permitted by statutory regulation or exceeds the permitted use, you will Underwood (Ed.), Eye guidance in reading and scene perception need to obtain permission directly from the copyright holder. To view a (pp. 29–53). Elsevier. copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . Kelly, D. J., Miellet, S., & Caldara, R. (2010). Culture shapes eye movements for visually homogeneous objects. Frontiers in Psy- chology, 1, 1–7. https:// doi. org/ 10. 3389/ fpsyg. 2010. 00006 Kim, S., Hasher, L., & Zacks, R. T. (2007). Aging and a benefit of References distractibility. Psychonomic Bulletin and Review, 14(2), 301–305. https:// doi. org/ 10. 3758/ BF031 94068 Angel, L., Bastin, C., Genon, S., Salmon, E., Fay, S., Balteau, E., King, J. A., Burgess, N., Hartley, T., Vargha-Khadem, F., & O’Keefe, J. Maquet, P., Luxen, A., Insingrini, M., & Collette, F. (2016). (2002). Human hippocampus and viewpoint dependence in spatial 1 3 Psychological Research (2022) 86:404–420 419 memory. Hippocampus, 12(6), 811–820. https://doi. or g/10. 1002/ Pertzov, Y., Miller, T. D., Gorgoraptis, N., Caine, D., Schott, J. M., But- hipo. 10070 ler, C., & Husain, M. (2013). Binding deficits in memory follow - Livingstone-Lee, S. A., Murchison, S., Zeman, P. M., Gandhi, M., ing medial temporal lobe damage in patients with voltage-gated van Gerven, D., Stewart, L., Livingston, N., & J. and Skelton, R. potassium channel complex antibody-associated limbic encepha- W. . (2011). Simple gaze analysis and special design of a virtual litis. Advance online publication. Brain: A Journal of Neurol- Morris water maze provides a new method for die ff rentiating ego - ogy,  136, 2474–2485.  https://doi.org/ https:// doi. or g/ 10. 1093/ centric and allocentric navigational strategy choice. Behavioural brain/ awt129 brain research, 225(1), 117–125. https:// doi. org/ 10. 1016/j. bbr. Postma, A., Kessels, R. P. C., & van Asselen, M. (2008). How the 2011. 07. 005 brain remembers and forgets where things are: The neurocogni- Luis, C. A., Keegan, A. P., & Mullan, M. (2009). Cross validation of tion of object-location memory. Neuroscience and Biobehavio- the Montreal Cognitive Assessment in community dwelling older ral Reviews, 32(8), 1339–1345. https://doi.org/https://doi. or g/10. adults residing in the Southeastern US. International Journal of 1016/j. neubi orev. 2008. 05. 001 Geriatric Psychiatry, 24(2), 197–201. https:// doi. org/ 10. 1002/ Raptis, G. E., Fidas, C. A., & Avouris, N. M. (2017). On implicit elici- gps. 2101 tation of cognitive strategies using gaze transition entropies in Mathôt, S., Schreij, D., & Theeuwes, J. (2012). OpenSesame: An pattern recognition tasks. In Proceedings of the 2017 CHI Con- open-source, graphical experiment builder for the social sciences. ference Extended Abstracts on Human Factors in Computing Sys- Behavior Research Methods, 44(2), 314–324. https:// doi. org/ 10. tems, 1993–2000. https://doi.org/https://d oi.o rg/1 0.1 145/3 02706 3. 3758/ s13428- 011- 0168-730531 06 Meulenbroek, O., Petersson, K. M., Voermans, N., Weber, B., & Ratcliff, R., Gomez, P., Thapar, A., & McKoon, G. (2004a). A diffusion Fernández, G. (2004). Age differences in neural correlates of route model analysis of the effects of aging in the lexical-decision task. encoding and route recognition. NeuroImage, 22(4), 1503–1514. Psychology and Aging, 19(2), 278–289. https:// doi. org/ 10. 1037/ https:// doi. org/ 10. 1016/j. neuro image. 2004. 04. 0070882- 7974. 19.2. 278 Moffat, S. D., Kennedy, K. M., Rodrigue, K. M., & Raz, N. (2007). Ratcliff, R., McKoon, G., & Gomez, P. (2004b). A diffusion model Extrahippocampal contributions to age differences in human spa- account of the lexical decision task. Psychological Review, 111(1), tial navigation. Cerebral Cortex, 17(6), 1274–1282. https:// doi. 159–182. https:// doi. org/ 10. 1037/ 0033- 295X. 111.1. 159 org/ 10. 1093/ cercor/ bhl036 Ratcliff, R., Thapar, A., & McKoon, G. (2006). Aging, practice, and Montefinese, M., Sulpizio, V., Galati, G., & Committeri, G. (2015). perceptual tasks: a diffusion model analysis. Psychology and Age-related effects on spatial memory across viewpoint changes Aging, 21(2), 353–371. https:// doi. org/ 10. 1037/ 0882- 7974. 21.2. relative to die ff rent reference frames. Psychological Research Psy - 353 chologische Forschung, 79(4), 687–697. https:// doi. org/ 10. 1007/ Reuter-Lorenz, P. A., & Cappell, K. A. (2008). Neurocognitive aging s00426- 014- 0598-9 and the compensation hypothesis. Current Directions in Psycho- Morcom, A. M., Li, J., & Rugg, M. D. (2007). Age effects on the neu- logical Science, 17(3), 177–182. https:// doi. org/ 10. 1111/j. 1467- ral correlates of episodic retrieval: increased cortical recruitment 8721. 2008. 00570.x with matched performance. Cerebral Cortex, 17(11), 2491–2506. Salthouse, T. A. (1996). The processing-speed theory of adult age dif- https:// doi. org/ 10. 1093/ cercor/ bhl155 ferences in cognition. Psychological Review, 103(3), 403. https:// Muffato, V., Hilton, C., Meneghetti, C., De Beni, R., & Wiener, J. M. doi. org/ 10. 1037/ 0033- 295x. 103.3. 403 (2019). Evidence for age-related deficits in object-location bind- Salthouse, T. A., & Ferrer-Caja, E. (2003). What needs to be explained ing during place recognition. Hippocampus, 29(10), 971–979. to account for age-related effects on multiple cognitive variables? https:// doi. org/ 10. 1002/ hipo. 23099 Psychology and Aging, 18(1), 91. https:// doi. org/ 10. 1037/ 0882- Nardini, M., Thomas, R. L., Knowland, V. C., Braddick, O. J., & Atkin-7974. 18.1. 91 son, J. (2009). A viewpoint-independent process for spatial reori- Schretlen, D., Pearlson, G. D., Anthony, J. C., Aylward, E. H., Augus- entation. Cognition, 112(2), 241–248. https:// doi. org/ 10.101 6/j. tine, A. M., Davis, A., & Barta, P. (2000). Elucidating the con- cogni tion. 2009. 05. 003 tributions of processing speed, executive ability, and frontal lobe Nasreddine, Z. S., Phillips, N. A., Bédirian, V., Charbonneau, S., volume to normal age-related differences in fluid intelligence. Whitehead, V., Collin, I., & Chertkow, H. (2005). The Montreal Journal of the International Neuropsychological Society, 6(1), Cognitive Assessment, MoCA: A brief screening tool for mild 52–61. https:// doi. org/ 10. 1017/ s1355 61770 06110 62 cognitive impairment. Journal of the American Geriatrics Soci- Segen, V., Avraamides, M. N., Slattery, T. J., & Wiener, J. M. (2020). ety, 53(4), 695–699. https:// doi. org/ 10. 1111/j. 1532- 5415. 2005. Age-related differences in visual encoding and response strategies 53221.x contribute to spatial memory deficits. Memory and Cognition. Nilakantan, A. S., Bridge, D. J., VanHaerents, S., & Voss, J. L. (2018). https:// doi. org/ 10. 3758/ s13421- 020- 01089-3 Distinguishing the precision of spatial recollection from its suc- Spaniol, J., Madden, D. J., & Voss, A. (2006). A diffusion model analy - cess: Evidence from healthy aging and unilateral mesial temporal sis of adult age differences in episodic and semantic long-term lobe resection. Neuropsychologia, 119, 101–106. https:// doi. org/ memory retrieval. Journal of Experimental Psychology: Learning 10. 1016/j. neuro psych ologia. 2018. 07. 035 Memory and Cognition, 32(1), 101–117. https://doi. or g/10. 1037/ Nuthmann, A. (2017). Fixation durations in scene viewing: Modeling 0278- 7393. 32.1. 101 the effects of local image features, oculomotor parameters, and Thapar, A., Ratcliff, R., & McKoon, G. (2003). A diffusion model task. Psychonomic Bulletin and Review, 24(2), 370–392. https:// analysis of the effects of aging on letter discrimination. Psychol- doi. org/ 10. 3758/ s13423- 016- 1124-4 ogy and Aging, 18(3), 415–429. https:// doi. org/ 10. 1037/ 0882- Olk, B., & Kingstone, A. (2015). Attention and ageing: Measuring 7974. 18.3. 415 effects of involuntary and voluntary orienting in isolation and Treitz, F. H., Heyder, K., & Daum, I. (2007). Differential course of in combination. British Journal of Psychology, 106(2), 235–252. executive control changes during normal aging. Aging, Neuropsy- https:// doi. org/ 10. 1111/ bjop. 12082 chology, and Cognition, 14(4), 370–393. https://doi. or g/10. 1080/ Pertzov, Y., Heider, M., Liang, Y., & Husain, M. (2015). Effects of 13825 58060 06784 42 healthy ageing on precision and binding of object location in vis- Verhaeghen, P., Cerella, J., & Basak, C. (2006). Aging, task com- plexity, and efficiency modes: the influence of working memory ual short-term memory. Psychology and Aging, 30(1), 26. https:// involvement on age differences in response times for verbal and doi. org/ 10. 1037/ a0038 396 1 3 420 Psychological Research (2022) 86:404–420 visuospatial tasks. Aging, Neuropsychology, and Cognition, 13(2), aging humans. Journal of Neuroscience, 33(14), 6012–6017. 254–280. https:// doi. org/ 10. 1080/ 13825 58909 69267https:// doi. org/ 10. 1523/ JNEUR OSCI. 0717- 12. 2013 Waldron-Perrine, B., & Axelrod, B. N. (2012). Determining an appro- Wilcox, R. R., & Keselman, H. J. (2003). Modem robust data analysis priate cutting score for indication of impairment on the Mon- methods: Measures of central tendency. Psychological Methods, treal Cognitive Assessment. International Journal of Geriatric 8(3), 254–274. https:// doi. org/ 10. 1037/ 1082- 989X.8. 3. 254 Psychiatry, 27(11), 1189–1194. https://doi. or g/10. 1002/ gps. 3768 Woods, D. L., Wyma, J. M., Yund, E. W., Herron, T. J., & Reed, B. Waller, D. (2006). Egocentric and nonegocentric coding in memory (2015). Age-related slowing of response selection and produc- for spatial layout: Evidence from scene recognition. Memory and tion in a visual choice reaction time task. Frontiers in Human Cognition, 34(3), 491–504. https://d oi.o rg/1 0.3 758/ BF0319 3573 Neuroscience, 9, 193. https://doi. or g/10. 3389/ fnhum. 2015. 00193 Waller, D. E., & Nadel, L. E. (2013). Handbook of spatial cognition. Zimmerman, S., Hasher, L., & Goldstein, D. (2011). Cognitive ageing: American Psychological Association. https:// doi. org/ 10. 1037/ a positive perspective. The paradoxical brain, 130–150 13936- 000 West, R. L. (1996). An application of prefrontal cortex function theory Publisher’s Note Springer Nature remains neutral with regard to to cognitive aging. Psychological Bulletin, 120(2), 272. https:// jurisdictional claims in published maps and institutional affiliations. doi. org/ 10. 1037/ 0033- 2909. 120.2. 272 Wiener, J. M., de Condappa, O., Harris, M. A., & Wolbers, T. (2013). Maladaptive bias for extrahippocampal navigation strategies in 1 3

Journal

Psychological ResearchSpringer Journals

Published: Mar 1, 2022

There are no references for this article.