Summary of Allen Newell's CHI '85 Address , "The Prospects for Science in Human-Computer Interaction " Thomas P . Moran Xerox PAR C The most popular highlight of' CHI '85 Conference wa s the Plenary Address by Professor Allen Newell o f Carnegie-Mellon University . His address was a ¢ frank and opinionated assessment of whether a psychological science of the user can have an impact on the field o f human-computer interaction -- not just by packing th e CHI conferences with psychologists, but by really contributing a science base that raises the quality of th e design of human-computer interfaces . By "science" Newell means a hard (quantitative . technical) science, not the flabby (qualitative . equivocal) science that psychology is usually though t to be . He summarized the competitive situatio n among sciences and disciplines by proposing an analo g of Gresham's Law : Hard science drives out sots . I f user psychology can't be hardened, then it will h e driven out of (actually, not let into) the field o f human-computer interaction by the hard engineerin g disciplines of computer science -- hardware . programming systems, computer graphics, etc . : eve n AI looks hard next to user psychology . Newell presented his vision of a hard user psycholog y having the following elements : ¢ It focuses on design, not evaluation . ¢ It has the form of an engineering-style theory based on task analysis, calculation, and approximation. ¢ It is used by interface designers at design time . This vision is developed in Card, Moran, and Newell' s 1983 book, The Psychology of Human-Compute r Interaction, which he used as a symbol of a genera l vision shared by a number of researchers . Newel l presented several models from the book as concrete , but partial, realizations of this vision . For example . the Model Human Processor is an architecture fo r integrating a collection of more specific models o f various cognitive functions . The model allows a wide variety of simple approximate calculations of huma n performance, and it thus provides the designer with a n effective way to think about the information - processing structure and behavior of the user . But this vision (as expressed in the book) has its critics . Newell took a step beyond the critics by succinctl y characterizing the problems with realizing the vision : ¢ The science is too low-level . ¢ Its scope is too limited . ¢ It is too slow to keep up with a rapidl y developing design field . ¢ It cannot be applied to real problems . He then discussed the prospects for dealing with eac h of these concerns . His message, basically, is that thi s approach is not the easy road ("the race is between th e tortoise of cumulative science and the hare of intuitiv e design") . For example, the Model Human Processo r provides us with a map of the state of the science : an d he laid out many of the areas where the model need s to be filled out. An especially interesting part of Newell's talk was hi s discussion of the levels issue . He presented a tabl e listing an exponential series of activity timescales, fro m 10- 3 seconds (milliseconds) to 10 9 seconds (decades) . He then asserted that the timescale of activit y determines the appropriate domain of science . Th e fastest timescales are governed by physical laws . Th e laws of human cognitive-symbolic system operate in a limited range of timescales (from tenths to tens o f seconds). Larger-scale behaviors (minutes to days t o years) are governed by laws of rationality ("commo n sense") or by laws of social interaction . Thus . it i s appropriate for cognitive psychology to focus at th e lower levels of user behavior . But the psychology als o affects the higher levels of rational behavior b y providing the analytic units . The enthusiastic response that Newell received to th e talk at CHI 85, along with Stu Card's volunteering t o share the authoring load, persuaded him to write it up . Their paper is scheduled to appear as a journal articl e in the 3rd issue this year of Human-Computer Interaction. SIGCHI Bulletin July 1985 Volume 17 Number 1

/lp/association-for-computing-machinery/summary-of-allen-newell-s-chi-85-address-the-prospects-for-science-in-xjnBqxbM40