TY - JOUR
AU - Orban, Guy A.
AB - Abstract In this study we used a modified double-label deoxyglucose procedure to investigate attention-dependent modulations of deoxyglucose uptake at the earliest stages of the macaque visual system. Specifically, we compared activity levels evoked during two tasks with essentially identical visual stimulation requiring different attentional demands. During a featural-attention task, the subjects had to discriminate the orientation of a grating; during a control spatial-attention task, they had to localize the position of a target point. Comparison of the resulting activity maps revealed attention-dependent changes in metabolic activity in portions of the magnocellular layers of the lateral geniculate nucleus, and the magnocellular-recipient layers 4Cα and 4B of the striate cortex. In these early stages of the visual system, attention to the orientation of the grating suppressed the metabolic activity in a retinotopically specific band peripheral to the representation of the stimulus. These results favor an early selection model of attention. After a thalamic attention-dependent gating mechanism, irrelevant visual information outside the focus of attention may be suppressed at the level of the striate cortex, which would then result in an increased signal-to-noise ratio for the processing of the attended feature in higher-tier, less retinotopically organized, extrastriate visual areas. Introduction At any given moment, a huge amount of information reaches our retinas. However, due to the restricted capacity of our brain to process information, only a fraction of this incoming retinal information can be adequately processed for behavioral purposes. One approach to reduce the information that needs to be processed is to attend selectively to one or a limited number of behaviorally important items or locations in our visual environment (Broadbent, 1958; Desimone and Duncan, 1995). When properties have to be identified from several simultaneously presented objects, subjects perform better and respond faster for attended (compared to non-attended) objects or positions in visual space (Posner, 1980; Treisman and Gelade, 1980; Van der Heijden et al., 1985 , 1988; Pan and Eriksen, 1993). Thus, selective attention to objects, features of objects, or locations in visual space seems to enhance the processing of attended items, and to reduce the processing of unattended items. This is confirmed by ample electrophysiological evidence showing that neuronal activity in many visual areas can be modulated by selective attention. For example, sensory responses of V2, V4, MT/V5 and MST neurons are modulated when monkeys attend to locations or features within the receptive field of these neurons (Connor et al., 1996, 1997; Treue and Maunsell, 1996; Haenny and Schiller 1988; Luck et al., 1997; McAdams and Maunsell, 1999). Human imaging studies have also provided evidence that selective featural and spatial attention can modulate cerebral blood flow in many visual areas. For example, it has been reported that attention to a particular attribute of an object enhances activity of these areas which are functionally specialized to process that particular attribute (Corbetta et al., 1991). In addition, paying attention to specific locations within the visual field yields increased blood flow at the corresponding representation within the retinotopically organized visual areas (Tootell et al., 1998; Watanabe et al., 1998a,b; Brefczynski and DeYoe, 1999; Martinez et al., 1999) and other extrastriate visual areas (Corbetta et al., 1993; Heinze et al., 1994; Nobre et al., 1997; Vandenberghe et al., 1997, 1996; Mangun et al., 1998). Despite clear evidence for attention-dependent modulation of neuronal activity in large portions of the extrastriate visual cortex, such modulation has been considered much weaker in striate cortex (Maunsell, 1995). However, because different investigators have reported conflicting results, the issue of attention-dependent modulations within the striate cortex remains unresolved. For example, Luck and co-workers observed no significant attention-dependent effects upon either spontaneous activity or sensory responses of V1 neurons (Luck et al., 1997). On the other hand, other authors reported moderate attention-dependent effects on the activity of about one-third of V1 neurons (Haenny and Shiller, 1988; Motter, 1993). The difficulty of investigating attention-dependent effects in V1, due to the small receptive fields of neurons in this area, may explain some of the discrepancies between the results obtained by different groups. Very small variations in eye position might also influence the activity of the neurons, thus interacting with potential attention-dependent effects. One way to circumvent problems inherent to eye movements is to apply very large stimuli so that minor changes in eye position do not covary with changes in retinal stimulation. Several recent human imaging studies adopted that strategy (Watanabe et al. 1998a,b; Brefczynski and DeYoe, 1999; Somers et al. 1999), and found profound attention-dependent modulation of activity in the striate cortex of human subjects. This is contrasted with another human imaging study in which only a very small positive spatial attention effect was observed in V1 by using a small visual stimulus (Tootell et al., 1998). Surprisingly, however, the latter study showed profound and widespread decreased activity in the early visual areas outside the representation of the stimuli which needed to be attended (Brefczynski and DeYoe, 1999; Somers et al., 1999). Thus, the question whether the activity of V1 neurons can or cannot be modulated by selective attention remains open and needs further investigation. To gain more insight into such attention-dependent changes in the early stages of the visual cortex, we applied an improved version (Tootell et al., 1988c; Geesaman et al., 1997) of the double-label deoxyglucose (DG) technique (Friedman et al., 1987, 1989; Redies et al., 1987) in the awake behaving monkey. This method offers the advantage that metabolic activity associated with two different physiological conditions can be isolated and compared within a single animal at high spatial resolution. This strategy makes it possible to detect attention-dependent effects over the whole extent of retinotopically organized visual areas, at the representation of the stimuli, and also in neighboring portions of the cortex. We intended to compare activity levels during two tasks which differed strongly in terms of visual attentional demands, in which visual stimulation and the behavioral response were as identical as possible. Specifically, we compared deoxyglucose uptake during execution of a featural-attention task to that during a spatial attention task. In the featural-attention task, the orientation of a grating was the cue controlling the monkey's behavior (Vogels and Orban, 1990). In the spatial attention task, the randomized position of a target point instead determined the monkey's response. Hence, in the latter task, the grating carried no behaviorally relevant information. Parts of this paper have been published previously in abstract form (Vanduffel et al., 1995). Materials and Methods Apparatus, Behavior and Surgical Procedures Five rhesus monkeys (Macaca mulatta) were used in this study. Four monkeys (M1–M4) were trained to perform the featural and spatial attention task, and one monkey (M5) served as control for the double-label DG procedures. Each monkey was restrained in a primate chair while its head was fixed with a stainless steel device, which was in turn fixed to the skull by means of orthopedic screws and dental cement. The monkeys were seated in the center of a pair of horizontally and vertically oriented coils which generated a magnetic field. Every 5 ms, eye positions were sampled using the magnetic scleral search coil technique (Judge et al., 1980). For this purpose, a Teflon-coated stainless steel coil was sutured on the sclera, which was subcutaneously routed to a connector in the headholder. The magnetically induced signals in the eye coil were filtered and amplified using an eye movement monitor (2i BD1, Indiran Instruments). The rhesus monkeys were trained in a completely dark room. All operations were performed under ketamine anesthesia (10 mg/kg Ketalar® i.m., Parke-Davis), supplemented with xylazine (0.5 mg/kg Rompun®, Bayer). During the eye coil surgery, anesthetics (0.2 mg Unicaine®, Bournonville Pharma), antibiotics (150 mg Lincocin®, Upjohn), corticosteroids (2 mg Celestone®, Schering-Plough) and a vasoconstrictor (5 μg Levorinine®, Sterop) were locally administered. Antibiotics (50 mg/kg i.m., Kefzol®, Lilly) were given daily during the week following each surgery. Three of the monkeys (M3–M5) received analgesics for 3 days after each surgery (5 mg/kg i.m. Tramadol, Dolzam®, Zambon). The surgical procedures conformed to American and European guidelines for the care and use of laboratory animals. The monkeys were water-deprived during the period of testing, and behavioral control was achieved using operant conditioning techniques. The health status of each monkey was closely monitored throughout the deprivation period, and monkeys were allowed to work until fully satiated. Furthermore, they received supplementary fruit during the weekends. Correct responses were rewarded by drops of apple juice, using a continuous reinforcement schedule. The use of saccadic eye movements as an operant enabled us to achieve a very high trial rate because of the very short latencies of the saccades. Eye position during fixation and saccadic eye movements towards the target points were detected by the computer using small windows centered on the fixation and target points. The size of the fixation windows were 0.7, 0.9, 1.4, 1.0 and 1.5 deg2 for M1, M2, M3, M4 and M5 respectively. Within each trial, the monkeys actually used only a tiny central portion of the fixation window. Featural-attention Task (Figs 1A, 2A) Description of Task We refer to this orientation identification task as a ‘featural’-attention task since the only cue for solving this task correctly is the orientation of a grating. In the featural-attention task, a fixation point appeared, and the monkeys had to begin fixating this point within 2 s (see Fig. 2A). After 100 ms of continuous fixation, a large circular square-wave grating (see below) appeared. Initially, the grating was tilted 35°, either clockwise or counterclockwise from vertical. This stimulus was presented foveally for 170 ms. Immediately after stimulus offset, two target points appeared to the left and the right side of the fixation point, at 8° eccentricity (see Figs 1A and 2A). The monkeys were required to make a saccadic eye movement towards the left target point when the grating was tilted counterclockwise, and towards the right target when the grating was tilted clockwise. When the monkey interrupted fixation prior to stimulus offset, the trial was aborted and a new one was started. After each saccade, an intertrial interval of 500 ms duration began. This featural-attention task is similar to the orientation identification task as previously used in human psychophysical and positron emission tomography studies (Vogels and Orban, 1986; Dupont et al., 1993; Vandenberghe et al., 1996; Orban et al., 1997). Training When the monkeys performed consistently at 85–90% correct with gratings tilted 35° from the vertical, we gradually decreased the orientation difference between the grating (relative to vertical), using a staircase procedure converging at a criterion of 84% correct. The orientation difference was decreased by 20% following correct responses in four successive trials, and the orientation difference was increased after each incorrect response. Every decrease or increase in orientation was stored as a reversal point. This staircase procedure enabled us to determine the just noticeable differences (JNDs) in orientation. The JNDs were calculated as the geometric mean of the reversal points within a session. On any given day, we initiated testing with a constant orientation difference which either matched the JND of the previous day, or was 20% smaller. After the subject again performed at 84% correct or better, the staircase procedure was introduced for the remaining of the session. On average, the subjects were trained for 43 days (on the featural-attention task) before they reached a stable performance level. During this period ~93 000 trials were completed. Final DG Session During the final DG session, two animals (M1 and M2) were tested using the staircase procedure. During this session, the orientation of the stimuli never deviated >2° from vertical. The third (M3) and the fourth (M4) monkey were tested using gratings which were tilted 5° and 2° clockwise or anticlockwise from vertical respectively. Spatial-attention Task (Figs 1B and 2B) When the subjects reached a stable performance level for the featural-attention task, we began the training for the spatial-attention task (see Fig. 1B). This task was a visually guided saccade paradigm in which a foveally presented grating must be ignored. We refer to this task as a ‘spatial’-attention task since the cue for solving the task correctly is the spatial position of a target point. The subjects need to divide their attention between two positions in the visual field, and they have to detect a target point appearing at one of the positions. Initially, an additional auditory cue was presented at the beginning of each trial to indicate the beginning of another task. In the spatial-attention task, the subject had to make a saccade towards a target point which appeared after stimulus offset, randomly positioned on the left or the right side of the fixation point (at 8° eccentricity). Behavioral performance in the spatial-attention task was measured using the percentage correct responses. A trial was considered correct if: (i) the subject did not interrupt fixation before completion of the trial, and (ii) when the subject finally made the saccade, it was towards the target point. When the subject made an incorrect saccade (e.g. an anti-saccade), this was considered an incorrect trial. When fixation was interrupted before the target appeared, the trial was aborted and a new one was started. On average, the monkey performed ~9000 trials during the training of the spatial-attention task. The trial rate for the two task conditions depended on the response latencies of the animal (see Table 1). As mentioned above, this rate was maximized (on average ~22 trials/min, excluding the aborted trials) and equalized between the two task conditions. The combination of the high trial rate and the excellent performance levels of the animals (JNDs in orientation of ~0.7° in the featural-attention task and ~97% correct responses in the spatial-attention task) ensured that the subjects attended either to the orientation of the grating or to the position of the target points. This also minimized possible differences in general arousal levels between the two tasks. Furthermore, it is very likely that the combination of a very high trial rate with excellent performance levels results in higher demands on the relevant neural substrates (Orban et al., 1997). Stimuli The stimuli used in the two task conditions were high-contrast (97%), square-wave gratings (0.5 c/deg for M1 and 1.0 c/deg for M2–M4) which were presented on a Brilliance 21A Phillips monitor (21 in., 70 Hz refresh rate) positioned 45 cm from the monkey's eyes. A noise pattern (20% white pixels) within the white stripes (3.0 cd/m2) resulted in equal ‘jagging’ of the grating edges at all orientations. The stationary gratings were presented foveally and confined within a circular electronic mask of different sizes (1.5°, 5°, 6° and 8° radius for M4, M3, M1 and M2 respectively). Since the experiments were performed in darkness, surrounding contours could not be used as cues for the discrimination. The phase of the grating was also randomized from trial to trial, to avoid the possibility that animals could use the position of the stripe ends to solve the task. There were neither contrast nor luminance differences between the two task conditions. Grating presentation time was equalized between the two conditions. Gratings were present for only 4.9% (M1), 5.9% (M2), 6.0% (M3) and 7.2% (M4) of time during the DG uptake period. Furthermore, the room was completely dark when the subjects were not fixating. Thus the amount of non-task related visual stimulation was strongly minimized. We were only interested in attention-dependent modulations of neuronal activity linked with the representation of the stimuli in the early visual areas and not in attention-dependent modulations of activity linked with decision or motor processes in the latest stages of the visual cortex or even non-visual cortical areas. Therefore, our prime concern was to present stimuli which were as identical as possible during the two tasks. There were only three minor differences between the tasks. (i) Only one target point was used in the spatial-attention task as opposed to two target points in the featural-attention task. However, this target point was presented on both sides for equal amounts of time. During the spatial attention task, the target points appeared at the same eccentricity as during the featural-attention task. Thus, the response rate, as well as the direction of the saccadic responses were the same during the two tasks. (ii) The behavioral response latencies differed slightly between tasks (see Table 1). To equalize trial rates between the two tasks, we uncoupled the stimulus on- and offset from the onset of the fixation and target points in the spatial-attention task, thereby compensating for differences in response latency (see Figs 1B and 2B). Thus, during the spatial-attention task, the grating does not carry behaviorally relevant information, either in the spatial or in temporal domains. (iii) Only a vertical grating was used in the spatial-attention task, to avoid possible ambiguities between stimulus orientation and position of the target point in this task (e.g. a counterclockwise-tilted grating paired with the right target point, or vice versa). The use of a vertical grating also minimizes the possibility that the monkey would covertly perform an featural-attention task during the spatial-attention task. It needs to be emphasized, though, that in the featural-attention task the stimuli never deviated >5° from vertical. Furthermore, for three monkeys (M1, M2 and M4) the difference from vertical never exceeded 2° (see above). Control Experiment (Fig. 3, M5) To clarify some methodological issues of the double-label DG technique (see below), one case of an additional control experiment was included (Vanduffel et al., 1997b). This control experiment illustrates that the monkeys fixate extremely well during the double-label DG experiments. Furthermore, it shows that the differences in DG uptake as observed in the present experiment cannot be explained by potential artifacts inherent to our procedure (see Discussion). In this control experiment, the monkey (M5) fixated passively for ~90% of the total experimental duration to a fixation point while we presented images of objects and scenes (Visual I, Fig. 3A) and scrambled versions of the same images (Visual II, Fig. 3B). Overlying the images, we presented a grid consisting of squares (2.2° × 2.2°) (see Fig. 3E). This grid corresponded to the borders of the scrambled parts and was present during both conditions. The monkey was rewarded only for fixation. The experimental procedures were exactly the same as described below for the spatial-attention and the featural-attention task, i.e. the experimental condition in which we showed the images of objects (visual I) was labeled with [3H]DG and the condition in which we showed the scrambled versions of the images (visual II) was labeled with [14C]DG. Double-label DG Experiment After the monkeys (M1–M4) had mastered both tasks, ~1 week before the DG experiment, a bilumen Deltacath® catheter (7 french × 15 cm, Becton Dickinson) was implanted in the external jugular vein, under general ketamine/xylazine anesthesia (see above). An incision was made in the skin and the external jugular vein exposed. Using scissors and a plastic tube, a subcutaneous channel was made from the headholder to the jugular vein. The catheter was inserted in the vein and secured with sutures. The distal (injection) side of the catheter was fixed to the headholder with dental cement. Using Luer locks, we could connect both ends of the bilumen catheter to secondary catheters which enabled us to make injections from outside the testing room without disturbing the monkeys. Every day preceding the final test session, the catheter was rinsed during the training session with 1.0 ml of a heparin solution (250 units/ml Heparine®, Rorer). In the final (DG) session, the monkeys were adapted to the featural-attention task until they reached a stable performance level (usually after ~5–20 min). Then, the [3H]DG was slowly (~1 min) injected (13.6–16.0 mCi/kg body wt; 40 Ci/mmol sp. act.; American Radiolabeled Chemicals, ARC). After 45 min of performance on the first task, the task was switched from featural-attention to spatial-attention. At that time, ~95% of the [3H]DG had been taken up from the bloodstream (Sokoloff et al., 1977). The monkeys were then allowed to adapt to the new task for 2 min before a bolus injection of [14C]DG (17–20 μCi/kg body wt; 55 mCi/mmol sp. act.; ARC) was administered through the second lumen of the catheter. The subject then performed the spatial-attention task for another 10 min. During the last 10 min of the featural-attention task and during the subsequent spatial-attention task, 3 ml heparin (5000 units/ml) was slowly administered. An overdose of sodium pentobarbital (100 mg/kg i.v.) was injected 10 min after the [14C]DG injection. The brevity of this second DG uptake period (10 min duration) prevents recirculation and/or loss of the first isotope. When stage 4 anesthesia was confirmed, the animal was perfused transcardially (for ~2–3 min) with phosphate-buffered saline containing 18% sucrose followed by a 1% paraformaldehyde fixative containing 1.5% glutaraldehyde and 18% sucrose. The perfusion ensured that residual unbound isotope was removed from the brain tissue. The brains were then rapidly removed. Striate cortex and parts of extrastriate cortex of one hemisphere was flattened and both hemispheres were frozen on precooled metal blocks lying on dry ice and stored thereafter at –80°C (Tootell and Silverman, 1985). Autoradiographic Procedures The flattened cortical tissue was cut in a cryostat (40 μm thick sections), parallel to the flattened cortical surface, at –20°C. The other hemisphere and the remainder of the first hemisphere was cryostat-sectioned in a horizontal plane. Sections were mounted on gelatin-coated (1%) coverslips and rapidly dried on a hotplate at ~65°C. The coverslips, together with 3H and 14C standards (Amersham), were pasted onto cardboard and exposed for 3–5 weeks against 3H-Hyperfilm (Amersham). The combination of the high 3H/14C ratio with the relatively short exposure time to the 3H-Hyperfilms produced an almost pure (see below) 3H signal on the Hyperfilm. Subsequently, the sections were exposed for ~3–4 months against Polaroid 891 color-positive films (Kronenberg, 1979). These films consist of three layers. The low-energy β-particles emitted by 3H (maximum β energy = 19 keV) excite only the outermost (blue) layer of this color-positive film. On the other hand, the higher-energy β-particles emitted by the 14C (maximum β energy = 156 keV), excite all three color-dye layers in the Polaroid film. Excitation of the Polaroid 891 film by the 14C yielded a pinkish-white image. A combination of several factors enabled a near-complete separation of the two activity linked maps. These factors included: (i) the very high 3H/14C ratio (800:1) relative to ratios used previously (Friedman et al., 1987); (ii) the shortened second DG uptake period and subsequent perfusion to remove unbound 14C; (iii) the use of color-positive Polaroid films, imaged through color filters (see below); and finally (iv) the short and long exposure times on the 3H-Hyperfilms and Polaroid films (respectively). Calibrations and in vivo tests showed that, quantitatively, the degree of isolation amounted to ~95% in the two films, without any image subtraction (R.B.H. Tootell, unpublished observations). After exposure, a selection of the sections was stained for Nissl and cytochrome oxidase activity. The different layers of area V1 were identified based upon the combination of their differential cytochrome oxidase reactivity and their differential baseline metabolic activity (Tootell et al., 1988a). The autoradiographs corresponding to the 3H signal from the Hyperfilm were digitized using a computerized imaging system (Analytical Imaging Concepts, Crosswell, 8-bit CCD camera). This image was first corrected by image subtraction of the background of the film in order to compensate for inhomogeneities of illumination and camera sensitivity. The alignment of the two functional images from one section was done very carefully using a micromanipulator which allowed fine translation and rotation adjustments. After digitizing the 3H image of one section (with a resolution of 520 × 480 pixels), this image was shown at inverted contrast and semi-transparency while the corresponding 14C image of the same section was also presented at half-transparency. This second ‘live’ image was aligned on-line with the first 3H image using features present in the isotope patterns, tissue border and tissue artifacts. Because of the opposite contrast between the two images, tissue artifacts and borders are equalized when the two images are perfectly aligned. When the two images were rotated relative to each other, this led to the so-called ‘glass pattern’ effects. When one functional image was translated relative to the other image, this resulted in a double-image. Spatial displacements >15 μm between the two functional images of the same section were detectable and could be corrected. After alignment, the 14C image was digitized through a yellow filter (Kodak Wratten Gelatin Filter no. 12) and this image was also corrected by image subtraction of the background. Using a similar procedure, we aligned 3H and 14C signals of consecutive sections. There is little distortion between images from adjacent cryostat sections because the tissue remains frozen in the period between sectioning and affixing to the coverslip. Therefore, we were able to average metabolic maps from two or three flattened sections, which increased image contrast and minimized artifacts from any one tissue section. Normalization Procedure A Gaussian filter (radius 10 μm) was applied to the 3H images in order to compensate for differences in photographic grain between the 3H and the 14C signals. Using calibrated 3H and 14C standards (Amersham), we plotted the normalized gray value–concentration curves of both isotopes, for each pair of films. For the gray values corresponding to neuronal activity, the gray value–concentration curves of both isotopes were shifted but had similar slopes. The constant shift between both sigmoids was added to the gray value of each pixel of the 3H autoradiographs to obtain a normalized 3H gray value. This normalization procedure, which is based on standards, does not assume equal 3H and 14C signals in the autoradiographs of the sections. Difference images (e.g. spatial-attention minus featural-attention) were scaled from –100 to +100 using the formula [(gray value 14C – normalized gray value 3H)/256 × 100]. For quantification, line plots of normalized isotope concentration were made between 0° and 8° eccentricity along the representation of the vertical and horizontal meridians in each of the V1 layers. From these line plots we extracted the data points corresponding to a particular eccentricity. Every sample was derived from a 50 × 50 μm2 wide region of cortex. Discrete measurements (50 × 50 μm2) were made at larger eccentricities and in the lateral geniculate nucleus (LGN) and the visual reticular thalamic nucleus (vRTN). In the LGN, the measurements were made in the middle of each layer (avoiding the borders of the layers) and they were spaced by ~250 μm along the representation of the horizontal meridian. All measurements were performed separately on the precisely aligned images of the 14C and normalized 3H signals (averaged over two or three consecutive sections to improve signal-to-noise ratio). Each data point, as discussed in the results, is derived from the average normalized isotope concentration of at least 45 measurements. Statistical significance was assessed using a two-tailed t-test. Visualization Procedures To visualize the results, we used several procedures. First, we show raw DG data. For example, in Figure 3A,B we present the individual 3H and 14C signals from a single V1 section of our control experiment in which we compared the DG uptake related to viewing images of objects and their scrambled versions. In this figure, we present data from a portion of the operculum from which the representation of the vertical meridians was removed (see Fig. 3F,G). In the striate cortex, one can observe virtually equal DG uptake related to viewing normal and scrambled versions of images. Furthermore, the figure shows nicely the retinotopic pattern corresponding to the grid which was present during both conditions. This proves that the monkey fixated extremely well during the experiment and that DG uptake related to very fine stimuli (i.e. the lines of the grid) can be traced in awake and behaving monkeys similar to that shown earlier in anaesthetized monkeys (Tootell et al., 1982, 1988b). We compared the two functional labels within one image in two ways. On the one hand, we show the subtraction image of the normalized 3H and 14C signals to enable a comparison between the two functional signals in one single image (e.g. Fig. 3C). For the subtraction image, we used a color-code in which blue signals equal DG uptake during the two task conditions and red means higher DG concentration in the [14C]DG signal (in this particular example; see color scalebar). In Figure 3C, virtually equal DG uptake can be observed in V1 after viewing images of objects and the scrambled versions of the same images. On the other hand, we used a combination image in which we put the inverted 3H signal in the red channel of an RGB image and the corresponding inverted 14C signal in the green channel. When the ‘blue’ channel (containing no image) is inverted, the latter procedure results in an image in which the two signals are differentially color-coded. When the two signals are equal, this results in a yellow image (as in Fig. 3D). When there is more [3H]DG than [14C]DG uptake, a green color is obtained, while the opposite ratio yields a reddish color. The latter presentation technique has the advantage over the subtraction technique in that the amplitude of both DG-signals can be visualized in the same image (e.g. compare Fig. 3C with 3D). Results Behavior The animals were highly overtrained for both tasks. Two monkeys (M1 and M2) worked at 84% correct during the final DG session, and their JNDs in orientation during that session were 0.68° and 0.70° respectively (see Table 1). Five days before the final DG session, the third animal (M3) reached a JND in orientation of 1.66° (see Table 1) and the fourth animal (M4) had a JND in orientation of 0.7°. However, in order to control for differences in task difficulty, we tested the latter two animals during the DG session using stimuli with a constant supra-threshold difference at an orientation of 5° (M3) and 2° (M4). By using this constant orientation difference during the featural-attention task, M3 and M4 responded correctly in 91% of the trials. In the spatial-attention task, the four animals responded correctly in 99, 96, 97, and 96% of the trials respectively (see Table 1). Metabolic Activity Overall Metabolic Activity In most areas of the brain, levels of the two labeled forms of DG uptake were essentially equal, after isotope normalization. In Figure 4, this is illustrated for area V1. Figure 4D shows normalized [3H]DG and [14C]DG concentration plots, which were measured from foveal to more peripheral visual field representations, along the line as indicated in the upper panels. The measurements were performed separately on precisely aligned images of the 14C and normalized 3H signals, which were averaged over two or three consecutive sections. This figure illustrates that baseline metabolic activity differs significantly between different layers, which is in agreement with previous DG studies (Kennedy et al., 1975; Hubel et al., 1977; Vanduffel et al., 1997a; Tootell et al., 1988a) and which is suggested by differential firing levels of neurons in different layers of the striate cortex (Snodderly and Gur, 1995). Furthermore, the normalized concentration plots through the different layers of V1 show that DG uptake during the two tasks was virtually identical in most of these layers (see Fig. 4D). Only in portions of layer 4Cα one can observe lower DG uptake related to the featural-attention task (see solid arrow in Fig. 4D). Finally, this figure shows that the difference in DG uptake is restricted to particular visual field representations within layer 4Cα: at foveal and parafoveal visual field representation DG uptake is virtually equal (see dotted arrow in Fig. 4D), while at more eccentric visual field representations one can observe profound differences in DG uptake. The equal DG uptake within most brain areas is in agreement with functional magnetic resonance imaging (fMRI) studies in the human showing that task-dependent modulation of neuronal activity occurs in only a small number of areas (Dupont et al., 1993; Vandenberghe et al., 1996). It must be emphasized, however, that the imaging techniques used in humans (PET and fMRI) and the optical imaging technique which can be used in monkeys do not reveal the layer-specific differential activity patterns revealed by the double-label DG technique. Striate Cortex Although we do observe layer-specific and retinotopic-specific DG uptake in the striate cortex, it remains unclear why such effects occur only in a restricted portion of layer 4Cα. Is the retinotopic position of the modulated DG uptake linked with the representation of the stimulus? This question can be answered by relating the observed modulation with the stimulus-driven DG uptake. Figure 5B shows the [3H]DG signal of a single section through layers 2–3 of the flattened operculum of M4. One can observe the higher stimulus-driven activity from the representation of the fovea to the representation of the border of the stimulus (at 1.5° eccentricity as indicated by the dotted line in Fig. 5A). At 8° azimuth along the representation of the horizontal meridian, one can also observe the representation of the target point. Despite the fact that we did observe stimulus-driven activation in all layers of the striate cortex of M4, the plots of normalized DG concentrations along the horizontal meridian (Fig. 5C), do not reveal task-dependent differential activity in layers 2–3. However, peripheral to the representation of the stimulus in layer 4Cα, one can observe lower DG uptake related to the featural-attention task (see Fig. 6). In the upper panel of Figure 6A, the 3H signal related to the featural-attention task is shown from a portion of a single section through V1. The middle panel shows the corresponding 14C signal related to the spatial-attention task. The red color in the pseudo-colored combination image (bottom panel of Fig. 6A) illustrates the lower featural-attention related DG uptake in isolated layer 4Cα just peripheral to the representation of the stimulus (as indicated by the red arrow in the upper panel). The yellow color in the bottom panel of Figure 6A signals equal DG uptake during the two task conditions. In Figure 6B, a similar color-code is used for an assembly of five sections to reconstruct nearly the complete extent of layer 4Cα οf the opposite hemisphere. The comparison of the stimulus -driven activity (as shown in Fig. 5B) with the lower DG uptake related to the featural-attention task (red in Fig. 6B), reveals that this lower DG uptake during the featural-attention task is present just peripheral to the representation of the stimulus. Moreover, Figure 6B shows that the lower featural-attention related DG uptake in layer 4Cα is not restricted to the representation of a principal meridian but includes all eccentricities peripheral to the representation of the stimulus. This finding argues against a potential mechanism for the differential DG uptake related to the representation of the target point at the representation of the horizontal meridian. Thus, in the subject which viewed the small gratings (M4), the differential DG uptake occurred as a ring surrounding the representation of the stimulus in magnocellular-recipient layer 4Cα of the striate cortex. We did observe differential (i.e. task-specific) DG uptake in the striate cortex of all four monkeys (M1–M4), although there was no evidence for stimulus-driven DG uptake in the early visual regions of these monkeys for which large gratings were presented (M1–M3). The reason for the lack of stimulus-driven DG uptake in these three monkeys which viewed large gratings might be attributed to surround suppression in V1 (Sillito and Jones, 1996). For the representation of the target point, the lack of metabolic activity might be due the large differences in response latency (see Table 2), and thus, large differences in the presentation time of the target points between M4 (25% of total experimental duration) and the three other monkeys (~7% of the total experimental duration). Figure 7 shows results of M1 in which a stimulus of 12° diameter was presented. Figure 7A,B shows raw DG signals, from a single V1 section, related to the featural-attention task and the spatial-attention task respectively. Figure 7C shows a difference image between a normalized 14C signal and 3H signal for a near-complete reconstruction of layer 4Cα (based upon three consecutive sections). The red color in the difference image indicates lower DG uptake during the featural-attention task, compared to DG uptake during the localization task. A comparison of these data with the known retinotopic organization of V1 (e.g. Fig. 3) reveals that the differential activity in layer 4Cα occurred just peripheral to the representation of the stimulus, similar to that in M4. The same modulation of DG uptake was present in M3, in which we used a 10° diameter stimulus (see Fig. 4). In retinotopically specific portions of magnocellular recipient layer 4Cα, DG uptake was lower when the monkey (M3) performed the featural-attention task than when the monkey performed the spatial-attention task. DG uptake differed in striate regions where the peripheral visual field is represented (see solid arrow at the right-hand side of the middle panel in Fig. 4D), but absent where the foveal and parafoveal visual fields are represented (see dotted arrow at the left-hand side of the middle panel in Fig. 4D). Thus, the ring of differential DG uptake outside the representation of the stimulus is layer and retinotopic specific, irrespective of the size of the stimulus and minor differences in performance levels. Quantitative data from the four monkeys confirm these results and reveal a general pattern. First, the differential DG uptake in layer 4Cα is primarily restricted to regions surrounding the representation of the grating along both the horizontal as well as the vertical meridian representations (Fig. 8). Secondly, the effect is consistent between monkeys despite variations in performance and stimulus diameter between subjects. As shown in Figure 9, the ring of suppression gradually moves to more eccentric visual field representations as the stimuli become larger (compare upper with lower panels in Fig. 9). A subsequent stage along the magnocellular stream, layer 4B of V1, was the only other layer in striate cortex which showed reduced and significant DG uptake related to the featural-attention task in all four subjects. However, differences here were much less striking than in layer 4Cα (Table 2). This may be related to the overall reduction in DG uptake seen in this layer (see Fig. 4B,D) (Tootell et al., 1988a). Control measurements revealed no consistent differential DG uptake in parvo-recipient layer 4Cβ (Table 2) which exhibits higher baseline DG uptake than layer 4Cα, nor in other V1 layers (layers II–III, IVA, V and VI; Table 2). Only in subject M4 did layer 4Cβ show significantly lower DG uptake related to the featural-attention task (see Fig. 10). Thus, in area V1, the modulation of DG uptake is retino-topically specific and confined to specific, magnocellular-dominated subdivisions. An important related question is whether this reflects either enhanced DG uptake during the spatial-attention task, or suppressed uptake during the featural-attention task. This question can be answered by comparing the ratio of DG uptake between layer 4Cα versus layer 4Cβ, at a range of eccentricities, within a single label (see Fig. 11). Importantly, we could not observe differences between the [14C]DG uptake ratios at different eccentricities in the spatial-attention task, suggesting that there is no intrinsic variation of DG uptake with eccentricity in area V1. Thus by comparing the ratio of normalized DG uptake between layer 4Cα versus layer 4Cβ at different eccentricities, it is possible to demonstrate absolute (i.e. suppressed or enhanced DG uptake in either condition) instead of relative differences in DG uptake between the two tasks. This comparison revealed that the location with maximal differential DG uptake (arrowheads in Fig. 9), coincides with the location showing the lowest [3H]DG uptake ratio (4Cα/4Cβ) related to the featural-attention task (arrowhead in Fig. 11). This result strongly suggests that the differential DG uptake reflects suppressed activation during the featural-attention task, because no difference in DG uptake was observed in parvo-recipient layer 4Cβ between the two tasks. The same conclusion can be drawn from the plots of normalized [3H]DG and [14C]DG concentration in Figure 4D. This figure shows virtual equal spatial-attention related DG uptake ([14C]DG) at central visual field representations (dotted arrow) compared to more peripheral visual field representations (solid arrow). On the other hand, the featural attention-related DG uptake ([3H]DG) is lower at higher eccentricities compared to more foveal representations. These results, together with these of the comparison of the DG uptake ratios between layer 4Cα and layer 4Cβ (see above), show that the differential DG uptake is caused by suppressed DG uptake during the featural-attention task and not by enhanced DG uptake during the spatial-attention task. Those authors who reported attention-dependent effects in V1 neurons (Haenny and Shiller, 1988; Motter, 1993; Roelfsema et al., 1998; Vidyasagar, 1998) always showed enhanced activity in neurons of which the receptive fields were covered by the stimulus. Is there any evidence in the present experiment for enhanced attention-dependent activity at the representation of the stimulus? Despite the pronounced attention-related effects outside the representation of the stimulus in the present experiment, there were no consistent attention-related effects at the representation of the stimulus itself. Only two of the four subjects (M3 and M4) showed significantly (two-tailed t-test, P < 0.05) higher DG uptake related to the featural-attention task in layer 4Cα at the representation of the stimulus. Although the two other subjects (M1 and M2) did show slightly higher DG uptake at the representation of the stimulus, this was not significant (P > 0.05). Other V1 layers of M4 tended to exhibit moderately higher DG uptake related to the featural-attention task at the foveal and parafoveal visual field representations (e.g. Fig. 10). However, in M4, only layers 4Cα, 4B and 4Cβ reached a significance level of P < 0.05. Thus compared to the lower DG uptake outside the representation of the stimulus, the positive featural-attention related effects within the representation of the stimulus were much smaller and varied between the subjects. A possible explanation for the apparent discrepancy between the present results and these electrophysiological results showing positive attention-dependent effects at the representation of the stimulus (Motter, 1993; Haenny and Shiller, 1988; Roelfsema et al., 1998; Vidyasagar, 1998) might be due to differences in stimulus size. This interpretation is supported by the fact that the observed suppression effects peripherally to the representation of the stimulus decrease with decreasing stimulus diameter (see Fig. 9). On the contrary, positive featural-attention related DG uptake within the representation of the stimulus was observed only in the two subjects (M3 and M4) that viewed the smallest gratings. Although plausible, such possible effects of stimulus size need to be interpreted with caution because of the limited number of subjects viewing small or large stimulus diameters in the present study. Alternatively, potential positive attention-related DG uptake might be overshadowed by possible stimulus-driven DG uptake at the representation of the stimulus. LGNd As noted above, differential DG uptake was observed in a layer of the striate cortex receiving thalamic input (layer 4Cα), but little or no differential DG uptake was observed in layers further removed from the input (Casagrande and Kaas, 1994). Thus it was of interest to look in detail at the DG uptake in the LGN itself. We observed differential DG uptake largely restricted to the magnocellular layers of the LGN in all four animals, which is in agreement with the pronounced effects restricted to magnocellular-recipient layer 4Cα in V1 (see Figs 12 and 13). The effect in the LGN was also retinotopic. Differential DG uptake was present in the ventral part of the LGN where the peripheral visual field is represented (Fig. 12, row II), but absent more dorsolaterally in the nucleus, where the foveal and parafoveal visual fields are represented (see Fig. 12, row I). Quantification of the differential isotope uptake in the magnocellular LGN layers (see Fig. 13) revealed a retinotopically specific band of differential DG uptake, located primarily peripheral to the representation of the stimulus. Although only a few subjects were available, there appeared to be a correlation between the magnitude of the modulated DG uptake in the LGN and stimulus size. That is, the larger stimuli evoked larger differential DG uptake between the two tasks (see Table 3 and Fig. 13). With respect to the differential DG uptake observed in the LGN, it is important to emphasize that neurons in the LGN are not prominently orientation selective (Thompson et al., 1994). Thus, the differential DG uptake cannot be attributed to the extremely small orientation differences in stimuli used during the two tasks. vRTN Thus, we observed retinotopically organized modulated DG uptake in the LGN as in the striate cortex. As far as we know, only layer 6 of V1 and the thalamic reticular nucleus gives rise to retinotopically organized feedback projections into the LGN (Mitrofanis and Guillery, 1993). Since no differential DG uptake was observed in layer 6 of the striate cortex, we looked in detail at the DG uptake in the visual reticular thalamic nucleus (vRTN), the other source of retinotopic input to the LGN. This structure displayed higher DG uptake during the featural-attention task (see Fig. 12, row I and Table 3). It must be stressed, however, that the effects in the vRTN are less pronounced than these in the LGN and area V1. For example, in the case with the small grating (M4), we measured only 3% modulated DG uptake in the RTN. One of the reasons for the small differences in DG uptake in the RTN might be related to the fact that averaged measurements from the complete extent of this structure were used. This was done since, as far as we know, no maps of the retinotopic organization of the RTN are available. Therefore, data from regions with and regions without modulated DG uptake might have been pooled, leading to artificially low estimates of differential DG uptake in the RTN. The opposite sign of the modulated DG uptake in the vRTN compared to that in V1 and the LGN is in agreement with the inhibitory nature of the feedback projections of this structure onto the LGN (Sherman, 1996). Discussion Summary Using the double-label DG technique, we compared activity levels at the early stages of the macaque visual system which were evoked during the performance of a featural-attention task with those measured during the performance of a spatial-attention task. Attention-dependent changes in metabolic activity were observed in the magnocellular layers of the LGN and magnocellular-recipient layers 4Cα and 4B of the striate cortex. The modulation occurred as a retinotopically specific band of suppressed DG uptake concentrated at retinotopic representations peripheral to that of the stimulus. Methodological Considerations Can the observed differential DG uptake be attributed to some sort of artifact such as (i) differences in task difficulty and arousal, (ii) small differences between stimuli, (iii) differences in microsaccades elicited in the two tasks, or (iv) the order of tasks and injections? For the following reasons, we believe these artifactual interpretations to be very unlikely. In two monkeys (M3 and M4), we equalized differences in task difficulty by presenting a constant suprathreshold orientation difference during the featural-attention task of 5° and 2° respectively. However, no systematic differences were observed between the results of the four monkeys (see Table 2 and Figs 9 and 13). Furthermore, the monkeys were highly overtrained, and trial rate was maximized in order to minimize any possible differences in arousal. Moreover, because the featural-attention related suppression of DG uptake in the early visual brain regions was retinotopically organized, there are no grounds to believe that it was caused instead by possible minor differences in task difficulty. If task difficulty or arousal was a major factor influencing neuronal activity at the level of V1, one would expect homogenous modulated DG uptake at all eccentricities of V1. The extremely small variations in stimulus orientation (only 0.7° for two monkeys) also cannot explain the observed results. The observed differential activity is not stimulus driven (see above), and LGN cells at least are not orientation selective (Thompson et al., 1994). In addition, it has been suggested that large saccades might decrease regional cerebral bloodflow in striate and extrastriate visual cortex (Paus et al., 1995; Brandt et al., 1998). Since we use fixation windows smaller than ~1 deg2, only microsaccades could influence DG uptake in the present experiment (Leopold and Logothetis, 1998). Since microsaccades occur normally at a frequency of 2–0.5 Hz, the frequency of microsaccades even decreases during brief fixations (Steinman et al., 1973; Winterson and Collewijn, 1976; Bridgeman and Palca, 1980). Thus, because the stimulus appeared only for 170 ms, the incidence of microsaccades in our study is extremely low. This was confirmed in two control monkeys which did not participate in the present experiment but which were trained to execute exactly the same tasks. We observed only 1.9 microsaccades/100 trials in one monkey and 3.3 microsaccades/100 trials in the other monkey, using the same criteria as Leopold and Logothetis (eye movements with a mean amplitude of ~10 arcmin and a duration of 10–20 ms) (Leopold and Logothetis, 1998). Moreover, in these monkeys we did not observe systematic differences in the number of microsaccades between the two tasks. Thus, neither saccades nor microsaccades can explain the results of the present experiment. Finally, it could be argued that the variation in fixation positions within the fixation window were systematically offset between the two tasks. Evidently, an offset in fixation can cause artifactual differential activity in the early visual areas. However, our results cannot be explained by such hypothetical differences in fixation position because: (i) we randomly varied the spatial phase of the large gratings, which counteracts any possible differences in eye position between the two tasks; and (ii) the featural-attention related ring of suppressed DG uptake in magnocellular components of the LGN and V1 can extend over >10 visual degrees (see Figs 8, 9, 10 and 13). Therefore, systematic differences in fixation between the two tasks, although theoretically possible, do not account for our observations since they cannot exceed 1.5°. An important question is whether the observed modulation might be due to the order of injections we used. In Figure 12, row III, control data are shown from the LGN of a an awake fixating monkey (M5) in which we compared DG uptake related to images (34° × 26°) of objects with DG uptake related to scrambled versions (16 × 12 squares) of the same images (Vanduffel et al., 1997b). Although the same order of injections was used as in the present experiment, no differential activity can be observed at the level of the LGN (Fig. 12, row III). In addition, changing the order of the task conditions several days before the final DG session (of the present experiment) yielded no difference in performance levels. This shows that there is no interaction between performance levels and task order. Together with results from other double-label DG experiments in which the order of experimental conditions was reversed (Schoups et al., 1995; Orban et al., 1997; Vanduffel, et al., 1997c), this indicates that neither the type of isotope, nor the order of isotope injection, nor the order of experimental conditions can explain the modulations that we observed in the present experiment. Thus the results of the present experiment cannot be attributed to differences in task difficulty and arousal, small differences between stimuli, differences in microsaccades elicited in the two tasks, or the order of tasks and injections. Instead, the differential DG uptake can be attributed to differences in attention. However, from the present experiments we cannot ascertain which aspects of attention led to the differential DG uptake. One interpretation is that the differential DG uptake is caused by different processing mechanisms and/or a different neuronal substrate subserving spatial and featural attention. Among the alternative explanations, it might be that a difference in size of the attentional spotlight between the two tasks has led to the differential DG uptake. Future experiments are needed to clarify the exact nature of the observed differential attention-dependent DG uptake. Comparison with Electrophysiology Although the baseline metabolic activity in the magnocellular LGN was higher than that in the parvocellular LGN, the reverse was true for magnorecipient layer 4Cα compared to parvo-recipient layer 4Cβ in V1 (see Figs 4, 7 and 12). However, we did observe differential DG uptake in the magnocellular LGN as well as in layer 4Cα of V1. Thus the present differences in isotope uptake cannot be attributed to possible artifactual differences in baseline activity. Since we do observe modulated DG uptake outside the representation of the stimulus, it is most likely that it acts upon the neurons which are spontaneously active. This could explain why such effects are much easier to detect in these layers showing higher spontaneous activity levels (e.g. layer 4Cα), compared to layers receiving considerable magnocellular input but exhibiting lower spontaneous activity levels such as the supragranular layers (e.g. Fig. 4) (Snodderly and Gur, 1995). Of course, we cannot decide whether or not visually driven activity is also suppressed by attention, since we did not present probe stimuli in the surround of the attended grating. It is not yet apparent why the magno-dominated system is affected by attention-dependent mechanisms in our experiment. This might be due to the relatively low spatial frequency of the stimuli or the fast behavioral responses, favoring short latency neurons (Nowak et al., 1995). An alternative interpretation is that the feature ‘orientation’ is primarily processed in the magno-dominated system, whether or not this feature is behaviorally important. This is also suggested by another series of double-label DG experiments in which monkeys were trained to fixate to gratings of orthogonal orientations (Schoups et al., 1995). These experiments, in which the monkeys had to fixate without performing a task, revealed interdigitated orientation columns in layer 4Cα of area V1 but not in layer 4Cβ (Livingstone and Hubel, 1984). In addition, the interdigitated orientation columns in area V2 are centered over the magno-dominated thick stripes. Virtual no interdigitated orientation columns could be observed in the thin stripes of area V2. These results demonstrate that neurons in layer 4Cα of area V1 and neurons in the thick stripes of area V2 are clustered according to their orientation preference. Conversely, neurons in layer 4Cβ of area V1 and the thin stripes of area V2 are not orientation selective, or they are clustered irrespective of their orientation preference. Thus, the parameter orientation is important for the functional anatomical organization of the magno-dominated subcompartments of areas V1 and V2, and, at least at the spatial resolution of the DG technique, not for that of the parvo-dominated subcompartments of these areas. Altogether, the results from these passive viewing experiments and from the present experiment, might indicate that the feature ‘orientation’ is preferentially processed in the magno-dominated subcompartments of the early stages of the visual system. Future experiments are needed to clarify whether a task using features tuned for the parvocellular system (e.g. a color-discrimination task instead of an orientation-discrimination task) could lead to a modulation of DG uptake in the parvo-dominated subcompartments of the early stages of the visual system (e.g. in the parvocellular LGN layers and layer 4Cβ). Such a mechanism in which functionally segregated subcompartments in the early stages of the visual system are modulated by attention could explain feature-specific, attention-related modulation of neuronal activity in extrastriate visual areas (Corbetta et al., 1991; O'Craven et al., 1997; Wojciulik et al., 1998). Comparison with Human Imaging and Psychophysics In striate cortex, the suppression of DG uptake is confined to specific, magnocellular-dominated subdivisions and is also retinotopically specific. A similar suppression of activity surrounding the representation of attended stimuli has been reported in somatosensory cortex in humans (Drevets et al., 1995) and rats (Chapin and Woodward, 1982). The human PET study revealed that attending to an impending tactile stimulus to the fingers produced decreased bloodflow in adjacent cortical zones representing the face. Also an fMRI study investigating the retinotopic organization of spatial attention (Tootell et al., 1998) yielded a profound suppression of activity in regions outside the representation of attended stimuli. These authors suggested that the results could reflect a suppression of eye movements since it is more difficult for the subjects to maintain fixation during a covert attention task than during a passive viewing condition. In the present experiment, the monkeys had to maintain fixation during the stimulus presentation of both task conditions, yet we observed reduced DG uptake related to the featural-attention task in the periphery of the stimulus. Thus, because it is equally difficult for the subjects to suppress eye movements during both tasks, it is unlikely that the suppression of eye movements is the major factor contributing to the decreased DG uptake. Furthermore, there is accumulating psychophysical evidence for a suppressive surround in which the discriminability of objects adjacent to an attended object is reduced (Cave and Zimmerman, 1997; Caputo and Guerra, 1998; Bahcall and Kowler, 1999; Tsotsos et al., 1999). Their results imply that rather than simply enhancing a selected region or object, attention controls local trade-offs in processing capacity, so that enhancement of one object or one location is accomplished at the expense of the immediate surround. It has also been suggested that the attentional field effects observed in a same– different letter comparison task represent the inhibition of irrelevant stimuli rather than a positive selection mechanism (Pan and Eriksen, 1993). Finally, by using the complexity theory and the minimum cost principal, Tsotsos' theoretical model predicts a strong inhibitory attentional beam surrounding the focus of attention at the earliest levels of the visual system (Tsotsos, 1990). Thus, the present results are in agreement with these human imaging as well as psychophysical and modeling studies. Moreover, our data suggest a thalamic origin for this attention-dependent suppression of activity, surrounding the representation of attended stimuli in the primary sensory cortex. Such a precortical gating mechanism could limit transmission of retinotopically specific, irrelevant sensory information to higher, less retinotopically organized cortical levels. In turn, such a mechanism might promote processing of signals carrying greater behavioral significance, i.e. those originating within the representation of the stimulus. Suppression might be the most effective manner for such a gating mechanism, since neurons at early levels are already activated to near maximal levels by sensory stimuli. This would leave little possibility for enhanced firing (Posner and Dehaene, 1994). This is exactly what we observed in the present experiments: profound suppression outside the stimulus and little or no enhanced activity within the representation of the stimulus. Possible Origin of the Suppression Effect At first sight, the most likely source of the attention-dependent modulation of neuronal activity in the LGN is layer VI of the striate cortex (Wilson, 1993). However, no differential DG uptake was observed in this layer. Since the vRTN showed relatively higher DG uptake for the featural-attention task (see Table 3 and Fig. 13A), it is conceivable that the differential DG uptake in the LGN instead arises from negative feedback originating in the vRTN. Indeed, it has been suggested that relay neurons in thalamic areas adjacent to active foci are inhibited through a feedback mechanism from the (v)RTN (Steriade et al., 1986). The vRTN is anatomically equipped to generate such retino-topically specific, task-dependent modulation since (i) it sends retinotopically organized projections to the LGN, and (ii) it receives projections from the LGN as well as from the forebrain and various brain stem nuclei which can be the source of top-down influences upon the LGN (Crick, 1984; Sherman and Koch, 1986; Mitrofanis and Guillery, 1993). Conclusion Our results suggest that attention-dependent modulation of neuronal activity is already present at the level of the LGN and V1. The reduction of transmitted signals outside behaviorally important active zones in the thalamus might balance the competition for processing capacity in less retinotopically organized extrastriate cortical areas, in favor of those features of the visual scene which are used to control the animal's behavior (Desimone and Duncan, 1995). Notes The authors thank M. De Paep for developing the software and C. Fransen for training of two animals. Special thanks are due to W. Spileers, R. Vogels and A. Schoups for their help with the surgeries. Furthermore, we are indebted to R. Vogels for his continous help during the experiments and valuable discussions of the results. We also appreciate the technical support of P. Kayenberg, G. Meulemans, Y. Celis and G. Vanparrijs. This work was supported by grants of the Human Science Frontiers Program, the National Research Council of Belgium (NFWO G3106.94), the Flemish Regional Ministry of Education (GOA 95/6), and the Queen Elisabeth Medical Foundation. W.V.D. is a postdoctoral research fellow of FWO-Flanders. Address correspondence to Wim Vanduffel, Laboratorium voor Neuro- en Psychofysiologie, Katholieke Universiteit Leuven, Campus Gasthuisberg, Herestraat 49, Leuven B-3000, Belgium. Email: Wim. Vanduffel@med.kuleuven.ac.be. Table 1 Reaction times (behavioral response latency),trial rate and behavioral performance   Latency (ms)  Trial rate/min  JND or % corr  Aborts/100 trials  Performance levels during the featural attention task were indicated by just noticeable differences (JND) in orientation (for M1, M2) and percent correct (% corr) responses (for M3 and M4). During the spatial attention task the performance of all subjects was measured by percent correct responses. The trial rate was kept as high as possible and the rate was equalized between the two task conditions. The number of aborts/100 trials corresponds to the number of trials which were interrupted before completion (e.g. because the monkey stopped fixating).  Featural attention          M1  274  16.7  0.68°  9  M2  194  20.9  0.70°  8  M3  258  20.9  91.0%  6  M4  586  25.7  91.0%  9  Spatial attention          M1  295  18.0  99.4%  15  M2  211  20.2  96%  9  M3  264  21.2  97.0%  8  M4  612  23.5  96.0%  13    Latency (ms)  Trial rate/min  JND or % corr  Aborts/100 trials  Performance levels during the featural attention task were indicated by just noticeable differences (JND) in orientation (for M1, M2) and percent correct (% corr) responses (for M3 and M4). During the spatial attention task the performance of all subjects was measured by percent correct responses. The trial rate was kept as high as possible and the rate was equalized between the two task conditions. The number of aborts/100 trials corresponds to the number of trials which were interrupted before completion (e.g. because the monkey stopped fixating).  Featural attention          M1  274  16.7  0.68°  9  M2  194  20.9  0.70°  8  M3  258  20.9  91.0%  6  M4  586  25.7  91.0%  9  Spatial attention          M1  295  18.0  99.4%  15  M2  211  20.2  96%  9  M3  264  21.2  97.0%  8  M4  612  23.5  96.0%  13  View Large Table 2 Percentage differential DG uptake in all layers of V1   2-3  4A  4B  4Cα  4Cβ  5  6  Normalized 3H and 14C concentrations were measured in 3–5 different sections for each animal. The measurements were averaged over these portions of V1 where the maximal differential DG uptake occurred. We averaged data from three eccentricities around this eccentricity where we observed the maximal differential DG uptake in layer 4Cα (as revealed by Fig. 9). For all layers, we chose the same eccentricities to obtain an average percent differential DG uptake. The brain regions from which samples were measured corresponded to the examples as illustrated in Figures 4, 6, 7 and 12. The percentage difference was calculated as ([featural attention] – [spatial attention])/[spatial attention] × 100. [featural attention] is the mean 3H concentration (related to the featural attention task) and [spatial attention] is the mean 14C concentration (related to the spatial attention task) for the respective structure.  M1  –2  –3  –2  –10  –2  2  –2  M2  1  –5  –2  –18  4  –2  –2  M3  –4  1  –3  –9  2  5  3  M4  1  –1  –3  –13  –5  –2  –2    2-3  4A  4B  4Cα  4Cβ  5  6  Normalized 3H and 14C concentrations were measured in 3–5 different sections for each animal. The measurements were averaged over these portions of V1 where the maximal differential DG uptake occurred. We averaged data from three eccentricities around this eccentricity where we observed the maximal differential DG uptake in layer 4Cα (as revealed by Fig. 9). For all layers, we chose the same eccentricities to obtain an average percent differential DG uptake. The brain regions from which samples were measured corresponded to the examples as illustrated in Figures 4, 6, 7 and 12. The percentage difference was calculated as ([featural attention] – [spatial attention])/[spatial attention] × 100. [featural attention] is the mean 3H concentration (related to the featural attention task) and [spatial attention] is the mean 14C concentration (related to the spatial attention task) for the respective structure.  M1  –2  –3  –2  –10  –2  2  –2  M2  1  –5  –2  –18  4  –2  –2  M3  –4  1  –3  –9  2  5  3  M4  1  –1  –3  –13  –5  –2  –2  View Large Table 3 Percentage differential DG uptake in the magnocellular layers of the LGN and vRTN   LGN  RTN  Same conventions as in Table 2.  M1  –32  21  M2  –30  33  M3  –17  12  M4  –11  3    LGN  RTN  Same conventions as in Table 2.  M1  –32  21  M2  –30  33  M3  –17  12  M4  –11  3  View Large Figure 1. View largeDownload slide  Experimental and stimulus design. (A) During the featural-attention task, the animals had to fixate for 100 ms before a large, high-contrast, circular square-wave grating appeared for 170 ms. Immediately after stimulus offset, two target points appeared to the left and the right of the fixation point at 8° eccentricity. The animals had to make a saccadic eye movement towards the left target when the grating was tilted counterclockwise, and towards the right target when the grating was tilted clockwise. (B) During the spatial attention task, the subjects had to make a saccade towards a target point which appeared after stimulus offset, positioned randomly to the left or the right side of the fixation point. The orientation difference in (A) has been exaggerated for clarity. In the featural-attention task the orientation of the grating never deviated >5° (clockwise or counterclockwise) from the vertical. For two of the animals, the difference never exceeded 2° (the just noticeable differences in orientation were 0.68° and 0.70°). Presentation time was equalized between the two conditions and amounted to only ~5% of the DG uptake period. Figure 1. View largeDownload slide  Experimental and stimulus design. (A) During the featural-attention task, the animals had to fixate for 100 ms before a large, high-contrast, circular square-wave grating appeared for 170 ms. Immediately after stimulus offset, two target points appeared to the left and the right of the fixation point at 8° eccentricity. The animals had to make a saccadic eye movement towards the left target when the grating was tilted counterclockwise, and towards the right target when the grating was tilted clockwise. (B) During the spatial attention task, the subjects had to make a saccade towards a target point which appeared after stimulus offset, positioned randomly to the left or the right side of the fixation point. The orientation difference in (A) has been exaggerated for clarity. In the featural-attention task the orientation of the grating never deviated >5° (clockwise or counterclockwise) from the vertical. For two of the animals, the difference never exceeded 2° (the just noticeable differences in orientation were 0.68° and 0.70°). Presentation time was equalized between the two conditions and amounted to only ~5% of the DG uptake period. Figure 2. View largeDownload slide  Time-course (in milliseconds) of stimulus presentation and responses during the featural-attention task (A) and the spatial-attention task (B). During the featural-attention task, two target points are presented, while during the spatial-attention task, only one target point appeared randomly on the left-or right-hand side of the grating. During the latter task, the stimulus on-set and off-set is uncoupled from fixation point and target point on-set. Figure 2. View largeDownload slide  Time-course (in milliseconds) of stimulus presentation and responses during the featural-attention task (A) and the spatial-attention task (B). During the featural-attention task, two target points are presented, while during the spatial-attention task, only one target point appeared randomly on the left-or right-hand side of the grating. During the latter task, the stimulus on-set and off-set is uncoupled from fixation point and target point on-set. Figure 3. View largeDownload slide  Raw DG data and comparison images of a control experiment in which we presented images of objects ([3H]DG signals in A) and the scrambled versions of the same images ([14C]DG signals in B). The raw DG signals are from a single section through a portion of the supragranular layers of flattened area V1. (C) The subtraction image between the normalized [14C]DG and the [3H]DG signals. Color scale (C): red = higher [14C]DG uptake; blue = no differential DG uptake. (D) A combination image of the [14C]DG and the [3H]DG signals (see Materials and Methods). (E) The grid which was presented together with the images of the objects as well as the scrambled images of the same objects. The retinotopic DG pattern which can be observed in (A) and (B) corresponds to the representation of this grid. (G) A diagram of the retinotopic organization (see F) of the portion (grey outline) from the operculum shown (black outline). Green = upper vertical meridian; blue = lower vertical meridian; red = horizontal meridian; orange = 6° eccentricity. Figure 3. View largeDownload slide  Raw DG data and comparison images of a control experiment in which we presented images of objects ([3H]DG signals in A) and the scrambled versions of the same images ([14C]DG signals in B). The raw DG signals are from a single section through a portion of the supragranular layers of flattened area V1. (C) The subtraction image between the normalized [14C]DG and the [3H]DG signals. Color scale (C): red = higher [14C]DG uptake; blue = no differential DG uptake. (D) A combination image of the [14C]DG and the [3H]DG signals (see Materials and Methods). (E) The grid which was presented together with the images of the objects as well as the scrambled images of the same objects. The retinotopic DG pattern which can be observed in (A) and (B) corresponds to the representation of this grid. (G) A diagram of the retinotopic organization (see F) of the portion (grey outline) from the operculum shown (black outline). Green = upper vertical meridian; blue = lower vertical meridian; red = horizontal meridian; orange = 6° eccentricity. Figure 4. View largeDownload slide  Plots of normalized [14C]DG and [3H]DG concentrations as a function of eccentricity in different layers of flattened area V1. (A–C) The 3H signals (related to the featural-attention task) from single V1 sections through layers 2–3, 3–4 and 5–6 respectively. No orientation columns or visually driven enhanced activity can be observed. In the three sections, the fovea is represented towards the left, with more peripheral visual field representations towards the right. The upper visual field is represented in the lower portion of the section. (D) Plots of normalized [3H]DG and [14C]DG concentrations as measured along the lines as indicated in (A–C) (along the representation of the horizontal meridian from foveal to more peripheral visual field representations). The different layers are differentially color-coded. Note the suppressed [3H]DG concentration in more peripheral (solid black arrow in D), but not foveal (dotted black arrow in D) visual field representations of layer 4Cα. Figure 4. View largeDownload slide  Plots of normalized [14C]DG and [3H]DG concentrations as a function of eccentricity in different layers of flattened area V1. (A–C) The 3H signals (related to the featural-attention task) from single V1 sections through layers 2–3, 3–4 and 5–6 respectively. No orientation columns or visually driven enhanced activity can be observed. In the three sections, the fovea is represented towards the left, with more peripheral visual field representations towards the right. The upper visual field is represented in the lower portion of the section. (D) Plots of normalized [3H]DG and [14C]DG concentrations as measured along the lines as indicated in (A–C) (along the representation of the horizontal meridian from foveal to more peripheral visual field representations). The different layers are differentially color-coded. Note the suppressed [3H]DG concentration in more peripheral (solid black arrow in D), but not foveal (dotted black arrow in D) visual field representations of layer 4Cα. Figure 5. View largeDownload slide  Stimulus-driven DG uptake in V1 of monkey M4. (B) The [3H]DG signal (featural-attention task) is shown of a section through layers 2–3 of flattened area V1. The representation of the grating is visible on the left-hand side of the section (the stimulus border is indicated by a dotted line in A). The representation of the target point is visible as a dark spot on the right-hand side of the section (B). (C) The plots of normalized DG uptake, as measured along the line as indicated in (B), show virtual equal DG uptake in layers 2–3 during the two task conditions. Note also the thick, thin, and interstripes in V2 (B) just beyond the representation of the vertical meridian (A). Figure 5. View largeDownload slide  Stimulus-driven DG uptake in V1 of monkey M4. (B) The [3H]DG signal (featural-attention task) is shown of a section through layers 2–3 of flattened area V1. The representation of the grating is visible on the left-hand side of the section (the stimulus border is indicated by a dotted line in A). The representation of the target point is visible as a dark spot on the right-hand side of the section (B). (C) The plots of normalized DG uptake, as measured along the line as indicated in (B), show virtual equal DG uptake in layers 2–3 during the two task conditions. Note also the thick, thin, and interstripes in V2 (B) just beyond the representation of the vertical meridian (A). Figure 6. View largeDownload slide  Ring of lower featural attention-related DG uptake surrounding the representation of the stimulus in layer 4Cα of V1 (M4). (A) The two upper panels show the raw [3H]DG and [14C]DG signals of a portion of a single section through flattened V1. The bottom panel shows the color-coded combination image of the two signals from isolated layer 4Cα. The fovea is represented to the left, and the border of the stimulus is indicated by the red arrow at the top (A). The green patch in the bottom panel (indicated by the blue asterix), is due to a small 3H-Hyperfilm artifact. (B) A similar color-coded combination image of layer 4Cα of the opposite hemisphere. Layer 4Cα is assembled from five consecutive sections. In (B), the stimulus border is indicated by the two red arrows (compare this border with the stimulus border as shown in Fig. 5B). Note in (A) and (B) a red ‘ring’ of lower featural-attention related DG uptake peripherally to the representation of the stimulus and more centrally to the representation of the target point. Color scale: red = lower featural attention-related DG uptake; yellow = no differential DG uptake. Figure 6. View largeDownload slide  Ring of lower featural attention-related DG uptake surrounding the representation of the stimulus in layer 4Cα of V1 (M4). (A) The two upper panels show the raw [3H]DG and [14C]DG signals of a portion of a single section through flattened V1. The bottom panel shows the color-coded combination image of the two signals from isolated layer 4Cα. The fovea is represented to the left, and the border of the stimulus is indicated by the red arrow at the top (A). The green patch in the bottom panel (indicated by the blue asterix), is due to a small 3H-Hyperfilm artifact. (B) A similar color-coded combination image of layer 4Cα of the opposite hemisphere. Layer 4Cα is assembled from five consecutive sections. In (B), the stimulus border is indicated by the two red arrows (compare this border with the stimulus border as shown in Fig. 5B). Note in (A) and (B) a red ‘ring’ of lower featural-attention related DG uptake peripherally to the representation of the stimulus and more centrally to the representation of the target point. Color scale: red = lower featural attention-related DG uptake; yellow = no differential DG uptake. Figure 7. View largeDownload slide  Ring of lower featural-attention related DG uptake surrounding the representation of the stimulus in layer 4Cα of V1 (monkey M1). (A,B) The raw [3H]DG and the [14C]DG signals of a single section through flattened area V1. The fovea is represented to the left. In this case no stimulus-driven DG uptake can be observed. Note the higher DG signals in layer 4Cα of (B) compared to (A) (e.g. compare the relative DG uptake between layers 4A and 4Cα in A and B). (C) The subtraction image between the 14C signal and the 3H signal for a reconstruction of layer 4Cα (based upon three consecutive sections). The normalized 3H gray value was subtracted from the 14C gray value on a pixel-by-pixel basis. Color scale (C): red = lower featural attention-related DG uptake; blue = no differential DG uptake. (D) The corresponding retinotopic organization of the operculum. Figure 7. View largeDownload slide  Ring of lower featural-attention related DG uptake surrounding the representation of the stimulus in layer 4Cα of V1 (monkey M1). (A,B) The raw [3H]DG and the [14C]DG signals of a single section through flattened area V1. The fovea is represented to the left. In this case no stimulus-driven DG uptake can be observed. Note the higher DG signals in layer 4Cα of (B) compared to (A) (e.g. compare the relative DG uptake between layers 4A and 4Cα in A and B). (C) The subtraction image between the 14C signal and the 3H signal for a reconstruction of layer 4Cα (based upon three consecutive sections). The normalized 3H gray value was subtracted from the 14C gray value on a pixel-by-pixel basis. Color scale (C): red = lower featural attention-related DG uptake; blue = no differential DG uptake. (D) The corresponding retinotopic organization of the operculum. Figure 8. View largeDownload slide  Quantification of differential DG uptake in layer 4Cα along the representations of the horizontal and vertical meridian in V1 (monkey M4). Normalized differential DG uptake (using the spatial attention task as baseline) is plotted as a function of eccentricity. Percentage differential DG uptake = [(the normalized DG concentration related to the featural attention task) – (the normalized DG uptake related to the spatial attention task)]/(the normalized DG uptake related to the spatial attention task) × 100. Each individual data point is derived from at least 45 normalized isotope concentration measurements. Note that the reduced featural-attention related DG uptake is not restricted to the horizontal meridian. Therefore, the differential DG uptake cannot be attributed to the unequal representation of the target point during the two tasks. Figure 8. View largeDownload slide  Quantification of differential DG uptake in layer 4Cα along the representations of the horizontal and vertical meridian in V1 (monkey M4). Normalized differential DG uptake (using the spatial attention task as baseline) is plotted as a function of eccentricity. Percentage differential DG uptake = [(the normalized DG concentration related to the featural attention task) – (the normalized DG uptake related to the spatial attention task)]/(the normalized DG uptake related to the spatial attention task) × 100. Each individual data point is derived from at least 45 normalized isotope concentration measurements. Note that the reduced featural-attention related DG uptake is not restricted to the horizontal meridian. Therefore, the differential DG uptake cannot be attributed to the unequal representation of the target point during the two tasks. Figure 9. View largeDownload slide  Normalized differential DG uptake in layer 4Cα of V1 as a function of eccentricity is plotted for all four subjects. The measurements along the horizontal and vertical meridians are averaged. The standard deviations are shown for monkey M3. Squares indicate the eccentricities for which the difference in isotope concentration reached a P-value of < 0.05; two-tailed t-test. The radius of the gratings which were presented in each experiment is indicated by the lines at the bottom of each panel. Note that the shift of lower featural-attention DG uptake from parafoveal to larger eccentricities is correlated with the size of the stimulus. Same conventions as in Figure 8. Figure 9. View largeDownload slide  Normalized differential DG uptake in layer 4Cα of V1 as a function of eccentricity is plotted for all four subjects. The measurements along the horizontal and vertical meridians are averaged. The standard deviations are shown for monkey M3. Squares indicate the eccentricities for which the difference in isotope concentration reached a P-value of < 0.05; two-tailed t-test. The radius of the gratings which were presented in each experiment is indicated by the lines at the bottom of each panel. Note that the shift of lower featural-attention DG uptake from parafoveal to larger eccentricities is correlated with the size of the stimulus. Same conventions as in Figure 8. Figure 10. View largeDownload slide  Normalized differential DG uptake as a function of eccentricity is plotted for all layers of M4. Data from the vertical and horizontal meridian are averaged. Same conventions as in Figure 8. Figure 10. View largeDownload slide  Normalized differential DG uptake as a function of eccentricity is plotted for all layers of M4. Data from the vertical and horizontal meridian are averaged. Same conventions as in Figure 8. Figure 11. View largeDownload slide  The differential DG uptake reflects suppressed DG uptake related to the featural attention task. For each label (spatial attention: [14C]DG, featural attention: [3H]DG), the DG uptake ratio between layer 4Cα and 4Cβ is plotted at three different eccentricity groupings for M1–M3. The middle group (5–9° eccentricity, indicated by the arrow) corresponds to the maximum differential DG uptake between the two tasks as shown in Figure 9 (see arrows in Fig. 9). The results of M4 are not included in the analysis because, due to the smaller gratings, the differential DG uptake (between the two tasks) in this subject was located at more central visual field representations compared to M1–M3. Figure 11. View largeDownload slide  The differential DG uptake reflects suppressed DG uptake related to the featural attention task. For each label (spatial attention: [14C]DG, featural attention: [3H]DG), the DG uptake ratio between layer 4Cα and 4Cβ is plotted at three different eccentricity groupings for M1–M3. The middle group (5–9° eccentricity, indicated by the arrow) corresponds to the maximum differential DG uptake between the two tasks as shown in Figure 9 (see arrows in Fig. 9). The results of M4 are not included in the analysis because, due to the smaller gratings, the differential DG uptake (between the two tasks) in this subject was located at more central visual field representations compared to M1–M3. Figure 12. View largeDownload slide  Differential DG uptake in the magnocellular layers of the LGN, concentrated peripheral to the representation of the grating (box outlined in black = present experiment). Columns A and B are color-coded images of autoradiographs from the same sections, produced by the [3H]DG (featural attention) and the [14C]DG (spatial attention) respectively. Each image represents the average of two consecutive sections and is normalized for differences in film processing (see Materials and Methods). The color scale (A and B) indicates normalized isotope concentration. Column C shows the difference images between the 14C and 3H signals of columns A and B (the normalized 3H gray value was subtracted from the 14C gray value on a pixel-by-pixel basis). Color scale (C): red = lower featural attention-related DG uptake; blue = no differential DG uptake; white = higher featural attention-related DG uptake. Row I: horizontal sections are shown through the LGN at a dorsal level including the foveal and parafoveal visual field representation. Row II: a more ventral level is shown, including the representation of the peripheral visual field. In the rightmost panel (D), estimations of retinotopic coordinates (A: azimuth; E: elevation) (Malpeli and Baker, 1975) are superimposed on cytochrome oxidase stains (Cytox) of the same sections shown in columns A–C of rows I and II. The magnocellular layers are surrounded by black outlines in panels A–C. The stimulus used in this case was 5° in radius (M3). Row III: control data (columns A–C as above). Row III shows a horizontal section through the ventral part of the LGN of a monkey participating in another double-label DG experiment. This monkey was required to fixate images of objects (visual I) or scrambled versions of the same images (visual II; see Materials and Methods). In this experiment, differential activity cannot be observed at the level of the LGN (row III, panel C; see Results). Figure 12. View largeDownload slide  Differential DG uptake in the magnocellular layers of the LGN, concentrated peripheral to the representation of the grating (box outlined in black = present experiment). Columns A and B are color-coded images of autoradiographs from the same sections, produced by the [3H]DG (featural attention) and the [14C]DG (spatial attention) respectively. Each image represents the average of two consecutive sections and is normalized for differences in film processing (see Materials and Methods). The color scale (A and B) indicates normalized isotope concentration. Column C shows the difference images between the 14C and 3H signals of columns A and B (the normalized 3H gray value was subtracted from the 14C gray value on a pixel-by-pixel basis). Color scale (C): red = lower featural attention-related DG uptake; blue = no differential DG uptake; white = higher featural attention-related DG uptake. Row I: horizontal sections are shown through the LGN at a dorsal level including the foveal and parafoveal visual field representation. Row II: a more ventral level is shown, including the representation of the peripheral visual field. In the rightmost panel (D), estimations of retinotopic coordinates (A: azimuth; E: elevation) (Malpeli and Baker, 1975) are superimposed on cytochrome oxidase stains (Cytox) of the same sections shown in columns A–C of rows I and II. The magnocellular layers are surrounded by black outlines in panels A–C. The stimulus used in this case was 5° in radius (M3). Row III: control data (columns A–C as above). Row III shows a horizontal section through the ventral part of the LGN of a monkey participating in another double-label DG experiment. This monkey was required to fixate images of objects (visual I) or scrambled versions of the same images (visual II; see Materials and Methods). In this experiment, differential activity cannot be observed at the level of the LGN (row III, panel C; see Results). Figure 13. View largeDownload slide  Quantification of differential DG uptake in the LGN. Same conventions as in Figure 8. Measurements in magnocellular LGN were taken along the horizontal meridian (HM). The standard deviations are shown for one monkey (M1). Squares indicate the eccentricities for which the difference in isotope concentration reached a P-value of <0.05; two-tailed t-test). The radius of the gratings used in each case is indicated by the lines at the bottom of each panel. Figure 13. View largeDownload slide  Quantification of differential DG uptake in the LGN. Same conventions as in Figure 8. Measurements in magnocellular LGN were taken along the horizontal meridian (HM). The standard deviations are shown for one monkey (M1). Squares indicate the eccentricities for which the difference in isotope concentration reached a P-value of <0.05; two-tailed t-test). The radius of the gratings used in each case is indicated by the lines at the bottom of each panel. References Bahcall DO, Kowler E ( 1999) Attentional interference at small spatial separations. Vis Res  39: 71–86. Google Scholar Brandt T, Bartenstein P, Janek A, Dieterich M ( 1998) Reciprocal inhibitory visual-vestibular interaction. Visual motion stimulation deactivates the parieto-insular vestibular cortex. Brain  121: 1749–1758. Google Scholar Brefczynski JA, DeYoe EA ( 1999) A physiological correlate of the ‘spotlight’ of visual attention. Nature Neurosci  2: 370–374. Google Scholar Bridgeman B, Palca J ( 1980) The role of microsaccades in high acuity observational tasks. Vis Res  20: 813–817. Google Scholar Broadbent DE (1958) Perception and communication. London: Pergamon Press. Google Scholar Caputo G, Guerra S ( 1998) Attentional selection by distractor suppression. Vis Res  38: 669–689. Google Scholar Casagrande VA, Kaas J (1994) The afferent, intrinsic, and efferent connections of primary visual cortex in primates. In: Cerebral cortex (Peters A, Rockland KS, eds), pp. 201–259. Google Scholar Cave KR, Zimmerman JM ( 1997) Flexibility in spatial attention before and after practice. Psychol Sci  8: 399–403. Google Scholar Chapin JK, Woodward DJ ( 1982) Somatic sensory transmission to the cortex during movement: phasic modulation over the locomotor step cycle. Exp Neurol  78: 670–684. Google Scholar Connor CE, Gallant JL, Preddie DC, Van Essen DC ( 1996) Responses in area V4 depend on the spatial relationship between stimulus and attention. J Neurophysiol  75: 1306–1308. Google Scholar Connor CE, Preddie DC, Gallant JL, Van Essen DC ( 1997) Spatial attention effects in macaque area V4. J Neurosci   17: 3201–3214. Google Scholar Corbetta M, Miezin FM, Dobmeyer S, Shulman GL, Petersen SE ( 1991) Selective and divided attention during visual discriminations of shape, color, and speed: functional anatomy by positron emission tomography. J Neurosci  11: 2383–2402. Google Scholar Corbetta M, Miezin FM, Shulman GL, Petersen SE ( 1993) A PET study of visuospatial attention. J Neurosci  13: 1202–1226. Google Scholar Crick F ( 1984) Function of the thalamic reticular complex: the searchlight hypothesis. Proc Natl Acad Sci USA  81: 4586–4590. Google Scholar Desimone R, Duncan J ( 1995) Neural mechanisms of selective visual attention. Annu Rev Neurosci  18: 193–222. Google Scholar Drevets WC, Burton H, Videen TO, Snyder AZ, Simpson-JR , Raichle ME ( 1995) Blood flow changes in human somatosensory cortex during anticipated stimulation. Nature  373: 249–252. Google Scholar Dupont P, Orban GA, Vogels R, Bormans G, Nuyts J, Schiepers C, De RM, Mortelmans L ( 1993) Different perceptual tasks performed with the same visual stimulus attribute activate different regions of the human brain: a positron emission tomography study. Proc Natl Acad Sci USA  90: 10927–10931. Google Scholar Friedman HR, Bruce CJ, Goldman RP ( 1987) A sequential double-label 14C-and 3H-2-DG technique: validation by double-dissociation of functional states. Exp Brain Res  66: 543–554. Google Scholar Friedman HR, Bruce CJ, Goldman RP ( 1989) Resolution of metabolic columns by a double-label 2-DG technique: interdigitation and coincidence in visual cortical areas of the same monkey. J Neurosci  9: 4111–4121. Google Scholar Geesaman BJ, Born RT, Andersen RA, Tootell RB ( 1997) Maps of complex motion selectivity in the superior temporal cortex of the alert macaque monkey: a double-label 2-deoxyglucose study. Cereb Cortex  7: 749–757. Google Scholar Haenny PE, Schiller PH ( 1988) State dependent activity in monkey visual cortex. I. Single cell activity in V1 and V4 on visual tasks. Exp Brain Res  69: 225–244. Google Scholar Heinze HJ, Mangun GR, Burchert W, Hinrichs H, Scholz M, Munte TF, Gos A, Scherg M, Johannes S, Hundeshagen H, et al. ( 1994) Combined spatial and temporal imaging of brain activity during visual selective attention in humans. Nature  372: 543–546. Google Scholar Hubel DH, Wiesel TN, Stryker MP ( 1977) Orientation columns in macaque monkey visual cortex demonstrated by the 2-deoxyglucose autoradiographic technique. Nature  269: 328–330. Google Scholar Judge SJ, Richmond BJ, Chu FC ( 1980) Implantation of magnetic search coils for measurement of eye position: an improved method. Vis Res  20: 535–538. Google Scholar Kennedy C, Des Rosiers M, Jehle JW, Reivich M, Sharp F, Sokoloff L ( 1975) Mapping of functional neural pathways by autoradiographic survey of local metabolic rate with 14C deoxyglucose. Science  187: 850–853. Google Scholar Kronenberg LH ( 1979) Radioautography of multiple isotopes using color negative films. Anal Biochem  93: 189–195. Google Scholar Leopold DA, Logothetis NK ( 1998) Microsaccades differentially modulate neural activity in the striate and extrastriate visual cortex. Exp Brain Res  123: 341–345. Google Scholar Livingstone MS, Hubel D ( 1984) Anatomy and physiology of a color system in the primate visual cortex. J Neurosci  4: 309–356. Google Scholar Luck SJ, Chelazzi L, Hillyard SA, Desimone R ( 1997) Neural mechanisms of spatial selective attention in areas V1, V2, and V4 of macaque visual cortex. J Neurophysiol  77: 24–42. Google Scholar Malpeli JG, Baker FH ( 1975) The representation of the visual field in the lateral geniculate nucleus of Macaca mulatta. J Comp Neurol  161: 569–594. Google Scholar Mangun GR, Buonocore MH, Girelli M, Jha AP ( 1998) ERP and fMRI measures of visual spatial selective attention. Hum Brain Map  6: 383–389. Google Scholar Martinez A, Anllo-Vento L, Sereno MI, Frank LR, Buxton RB, Dubowitz DJ, Wong EC, Hinrichs H, Heinze HJ, Hillyard SA ( 1999) Involvement of striate and extrastriate visual cortical areas in spatial attention. Nature Neurosci  2: 364–369. Google Scholar Maunsell JH ( 1995) The brain's visual world: representation of visual targets in cerebral cortex. Science  270: 764–769 [review]. Google Scholar McAdams CJ, Maunsell JR ( 1999) Effects of attention on orientation-tuning functions of single neurons in macaque cortical area V4. J Neurosci  19: 431–441. Google Scholar Mitrofanis J, Guillery RW ( 1993) New views of the thalamic reticular nucleus in the adult and the developing brain. Trends Neurosci  16: 240–245. Google Scholar Motter BC ( 1993) Focal attention produces spatially selective processing in visual cortical areas V1, V2, and V4 in the presence of competing stimuli. J Neurophysiol  70: 909–919. Google Scholar Nobre AC, Sebestyen GN, Gitelman DR, Mesulam MM, Frackowiak RS, Frith CD ( 1997) Functional localization of the system for visuospatial attention using positron emission tomography. Brain  120: 515–533. Google Scholar Nowak LG, Munk MH, Girard P, Bullier J ( 1995) Visual latencies in areas V1 and V2 of the macaque monkey. Vis Neurosci  12: 371–384. Google Scholar O'Craven KM, Rosen BR, Kwong KK, Treisman A, Savoy RL ( 1997) Voluntary attention modulates fMRI activity in human MT-MST. Neuron  18: 591–598. Google Scholar Orban GA, Dupont P, Vogels R, Bormans G, Mortelmans L ( 1997) Human brain activity related to orientation discrimination tasks. Eur J Neurosci  9: 246–259. Google Scholar Pan K, Eriksen CW ( 1993) Attentional distribution in the visual field during same–different judgments as assessed by response competition. Percept Psychophys  53: 134–144. Google Scholar Paus T, Marrett S, Worsley KJ, Evans AC ( 1995) Extraretinal modulation of cerebral blood flow in the human visual cortex: implications for saccadic suppression. J Neurophysiol  74: 2179–2183. Google Scholar Posner MI ( 1980) Orienting of attention. Quart J Exp Psychol  32: 3–25. Google Scholar Posner MI, Dehaene S ( 1994) Attentional networks. Trends Neurosci  17: 75–79. Google Scholar Redies C, Diksic M, Evans AC, Gjedde A, Yamamoto YL ( 1987) Double-label autoradiographic deoxyglucose methd for sequential measurement of regional cerebral glucose utilization. Neuroscience  22: 601–619. Google Scholar Roelfsema PR, Lamme VA, Spekreijse H ( 1998) Object-based attention in the primary visual cortex of the macaque monkey. Nature  395: 376–381. Google Scholar Schoups AA, Tootell RBH, Vanduffel W, Orban GA ( 1995) Use of the double-label deoxyglucose approach to map the orientation columnar system beyond area V1 and V2 in the macaque, and its plasiticity. Soc Neurosci Abstr  21: 15. Google Scholar Sherman SM ( 1996) Dual response modes in lateral geniculate neurons: mechanisms and functions. Vis Neurosci  13: 205–213. Google Scholar Sherman SM, Koch C ( 1986) The control of retinogeniculate transmission in the mammalian lateral geniculate nucleus. Exp Brain Res  63: 1–20. Google Scholar Sillito AM, Jones HE (1996) Context-dependent interactions and visual processing in V1. J Physiol (Paris) 90:205–209. Google Scholar Snodderly DM, Gur M ( 1995) Organization of striate cortex of alert, trained monkeys (Macaca fascicularis): ongoing activity, stimulus selectivity, and widths of receptive field activating regions. J Neurophysiol  74: 2100–2125. Google Scholar Sokoloff L, Reivich M, Kennedy C, Des RM, Patlak CS, Pettigrew KD, Sakurada O, Shinohara M ( 1977) The [14C]deoxyglucose method for the measurement of local cerebral glucose utilization: theory, procedure, and normal values in the conscious and anesthetized albino rat. J Neurochem  28: 897–916. Google Scholar Somers DC, Dale AM, Seiffert AE, Tootell RBH ( 1999) Functional MRI reveals spatially specific attentional modulation in human primary visual cortex. Proc Natl Acad Sci USA  96: 1663–1668. Google Scholar Steinman RM, Haddad GM, Skavenski AA, Wyman D ( 1973) Miniature eye movement. Science  181: 810–819. Google Scholar Steriade M, Domich L, Oakson G ( 1986) Reticularis thalami neurons revisited: activity changes during shifts in states of vigilance. J Neurosci  6: 68–81. Google Scholar Thompson KG, Zhou Y, Leventhal AG ( 1994) Direction-sensitive X and Y cells within the A laminae of the cat's LGNd. Vis Neurosci  11: 927–938. Google Scholar Tootell RB, Silverman MS, Switkes E, De Valois RL ( 1982) Deoxyglucose analysis of retinotopic organization in primate striate cortex. Science  218: 902–904. Google Scholar Tootell RB, Silverman MS ( 1985) Two methods for flat-mounting cortical tissue. J Neurosci Methods  15: 177–190. Google Scholar Tootell RB, Hamilton SL, Silverman MS, Switkes E ( 1988) Functional anatomy of macaque striate cortex. I. Ocular dominance, binocular interactions, and baseline conditions. J Neurosci  8: 1500–1530. Google Scholar Tootell RB, Swikes E, Silverman MS, Hamilton SL ( 1988) Functional anatomy of macaque striate cortex. Retinotopic organization. J Neurosci  8: 1531–1568. Google Scholar Tootell RBH, Born RT, Hamilton SL ( 1988) Studies of primate visual cortex using a double-label DG technique and color autoradiography. Soc Neurosci Abstr  14: 897. Google Scholar Tootell RB, Hadjikhani N, Hall EK, Marrett S, Vanduffel W, Vaughan JT, Dale AM ( 1998) The retinotopy of visual spatial attention. Neuron  21: 1409–1422. Google Scholar Treisman AM, Gelade G ( 1980) A feature-integration theory of attention. Cogn Psychol  12: 97–136. Google Scholar Treue S, Maunsell JH ( 1996) Attentional modulation of visual motion processing in cortical areas MT and MST. Nature  382: 539–541. Google Scholar Tsotsos JK ( 1990) Analyzing vision at the complexity level. Behav Brain Sci  13: 423–469. Google Scholar Tsotsos JK, Culhane S, Cutzu F (1999) From theortical foundations to a hierarchical circuit for selective attention. In :Visual attention and cortical circuits (Braun J, Koch C, eds). Cambridge, MA: MIT Press (in press). Google Scholar Van der Heijden AHC, Schreuder R, Wolters G ( 1985) Enhancing single-term recognition accuracy by cueing spatial locations in vision. Quart J Exp Psychol  A37: 427–434. Google Scholar Van der Heijden AHC, Wolters G, Enkeling M ( 1988) The effects of advance location cueing on latencies in a single-letter recognition task. Psychol Res  50: 94–102. Google Scholar Vandenberghe R, Dupont P, De BB, Bormans G, Michiels J, Mortelmans L, Orban GA ( 1996) The influence of stimulus location on the brain activation pattern in detection and orientation discrimination. A PET study of visual attention. Brain  119: 1263–1276. Google Scholar Vandenberghe R, Duncan J, Dupont P, Ward R, Poline JB, Bormans G, Michiels J, Mortelmans L, Orban GA ( 1997) Attention to one or two features in left or right visual field: a positron emission tomography study. J Neurosci  17: 3739–3750. Google Scholar Vanduffel W, Tootell RBH, Orban GA ( 1995) Metabolic mapping of an orientation discrimination task in the macaque using the double-label deoxyglucose technique. Soc Neurosci Abstr  21: 311. Google Scholar Vanduffel W, Vandenbussche E, Singer W, Orban GA ( 1997) A metabolic mapping study of orientation discrimination and detection tasks in the cat. Eur J Neurosci  9: 1314–1328. Google Scholar Vanduffel W, Vogels R, Tootell RBH, Orban GA ( 1997) Scrambling images of natural objects: II. A double-label deoxyglucose study. Invest Ophthalmol Vis Sci  38: S1001. Google Scholar Vanduffel W, Tootell RBH, Orban GA ( 1997) Macaque visual areas involved in color processing: a double-label deoxyglucose study. Soc Neurosci Abstr  23: 845. Google Scholar Vidyasagar TR ( 1998) Gating of neuronal responses in macaque primary visual cortex by an attentional spotlight. NeuroReport  9: 1947–1952. Google Scholar Vogels R, Orban GA ( 1986) Decision processes in visual discrimination of line orientation. J Exp Psychol Hum Percept Perform  12: 115–132. Google Scholar Vogels R, Orban GA ( 1990) How well do response changes of striate neurons signal differences in orientation: a study in the discriminating monkey. J Neurosci  10: 3543–3558. Google Scholar Watanabe T, Sasaki Y, Miyauchi S, Putz B, Fujimaki N, Nielsen M, Takino R, Miyakawa S ( 1998) Attention-regulated activity in human primary visual cortex. J Neurophysiol  79: 2218–2221. Google Scholar Watanabe T, Harner AM, Miyauchi S, Sasaki Y, Nielsen M, Palomo D, Mukai I ( 1998) Task-dependent influences of attention on the activation of human primary visual cortex. Proc Natl Acad Sci USA  95: 11489–11492. Google Scholar Wilson JR ( 1993) Circuitry of the dorsal lateral geniculate nucleus in the cat and monkey. Acta Anat  147: 1–13. Google Scholar Winterson BJ, Collewijn H ( 1976) Microsaccades during finely guided visuomotor tasks. Vis Res  16: 1387–1390. Google Scholar Wojciulik E, Kanwisher N, Driver J ( 1998) Covert visual attention modulates face-specific actvity in the human fusiform gyrus: fMRI study. J Neurophysiol  79: 1574–1578. Google Scholar © Oxford University Press
TI - Attention-dependent Suppression of Metabolic Activity in the Early Stages of the Macaque Visual System
JF - Cerebral Cortex
DO - 10.1093/cercor/10.2.109
DA - 2000-02-01
UR - https://www.deepdyve.com/lp/oxford-university-press/attention-dependent-suppression-of-metabolic-activity-in-the-early-5ndFLVvM3U
SP - 109
EP - 126
VL - 10
IS - 2
DP - DeepDyve
ER -