Editor’s Introduction to Visions of Cognitive Science

Visions of Cognitive Science will be an ongoing topic that is represented in this first issue by six extraordinary papers. Deciding on whom to invite for this first set of Visions papers was an effort that involved many people in the cognitive science community and was begun before we had completed contract negotiations with Wiley-Blackwell. As you will see when you read the papers, each author took our challenge to write an integrative and reflective paper very differently, and each has produced an interesting work that should stimulate thought and controversy across many of the disciplines of cognitive science. (And of course, topiCS is the right journal in which to publish a response to these authors.)

James McClelland leads this section with a very personal view of The Place of Modeling in Cognitive Science. In this wide-ranging essay, McClelland discusses why we model, what models can tell us, and how the agenda of cognitive modeling contrasts with that of artificial intelligence or machine learning. McClelland then provides a very ecumenical discussion of the strengths and contributions of the modeling paradigms that have developed over cognitive science’s first 30 years. He closes his essay with interesting speculation as to where each of these approaches might lead in the next 30 years as the power of our theories and computers continues to increase.

Karalyn Patterson and David Platt warn us that “Shallow Draughts Intoxicate the Brain” to draw Lessons from Cognitive Science for Cognitive Neuropsychology. Cognitive neuropsychology is that part of cognitive neuroscience that takes “as its evidential base, patterns of preserved and disrupted cognitive abilities arising from brain injury or disease” (Patterson & Plaut, 2009, this issue). The authors argue that mainstream cognitive neuropsychology has divorced itself from the intellectual and theoretical currents of mainstream cognitive science by unfortunate choices in data selection, data analysis, and theoretical goals. They then offer several prescriptions for bringing cognitive neuropsychology back into the mainstream. For a nonspecialist, the paper provides an interestingly written update on the field that used to provide the main source of neurocognitive data. We do not know how the mainstream cognitive neuropsychology community will react to Patterson and Plaut’s complaints and prescriptions, but we invite them to comment on these issues in a future issue of topiCS.

Turnabout is fair play and the paper by Nicola Clayton, James Russell, and Anthony Dickinson is a strike at mainstream cognitive psychology from one of cognitive science’s oldest, most revered, and arguably, most ignored subdisciplines; namely, animal cognition. Clayton and colleagues take aim at the mental time travel hypothesis (Roberts, 2002; Suddendorf & Corballis, 1997), which claims that animals cannot “recall past experiences (episodic memory) or anticipate future states (future planning) because they cannot dissociate themselves from the more or less immediate present” (Clayton, Russell, & Dickinson, 2009, this issue) and are therefore “stuck in time.” Through a series of clever studies (theirs and others), they demonstrate that corvids (members of the crow family) and nonhuman primates are capable of “recollecting the past and planning for the future.” They argue that the assumption to the contrary rests on “phenomenological assumptions based on human-centric ways of thinking,” which have driven a wedge between psychological studies of cognition and other approaches such as comparative studies (i.e., nonhuman animals) and artificial intelligence approaches.

In our third paper, rather than attacking another subdiscipline from the mainstream or attacking the mainstream from a subdiscipline, author Michelene Chi launches a frontal assault on her own subdiscipline, the Cognitive and Learning Sciences. Chi notes that whereas the terms “active, constructive, and interactive” are widely used to describe activities undertaken by learners, that the literature is “not explicit about how these terms can be defined; whether they are distinct; and whether they refer to overt manifestations, learning processes, or learning outcomes.” Arguing that a focus on overt learner activities is superior to the current jumble, she provides a taxonomy that distinguishes the three terms and that generates a testable hypothesis. For those who do not have a dog in this fight, Chi’s article provides an interesting and readable snapshot of the state of the art in the Cognitive and Learning Sciences. We predict that her colleagues will find Chi’s taxonomy to be a bold reformulation that will be debated and discussed for years to come.

Gerd Gigerenzer and Henry Brighton give us Homo Heuristicus, an extended argument on Why Biased Minds Make Better Inferences. There is much in this article to like and to recommend. First, there is a succinct reprise of the arguments advanced on both sides of Gigerenzer’s nearly 20-year fight against what was then the mainstream cognitive science perspective on decision making (Kahneman, 2003; Tversky & Kahneman, 1974). Second, there is the evolution of Gigerenzer’s adaptive toolbox as an ecologically rational adaption to the task environment. The bigger issue in Gigerenzer and Brighton’s view is not whether humans use the heuristics of the adaptive toolbox, but how they decide in what environment to use which heuristic. Another interesting development is the increasingly formal, model-based perspective that Gigerenzer and Brighton advance in their arguments against their opposition. Finally, the dominant chord throughout is a very interesting frontal assault on the prevailing view that more information, performing more computation, or taking more time is always better. The core of what is new and interesting here is the demonstration that, in many task environments less information, defined as ignoring cues, weights, and dependencies between cues can formally be shown to produce better solutions than more cues, more carefully refined weights, or a more appropriate accounting of dependencies between cues.

It seems appropriate to let Gary Marcus close the first Visions section of topiCS with what many of us would consider the ultimate vision for cognitive science, How Does the Mind Work? Marcus worries that the big foundational questions that formed the basis of many heated battles over the past 30 or more years have given way to a generation of researchers who are so lost in the details of their work as to not see the forest for the trees. The paper provides an historic background and update for four foundational issues: development (role of genes vs. experience), modularity, optimality, and symbol manipulation. As Marcus (2009, this issue) says, “To move further, we need to know not just the general sorts of things the brain does (e.g., processing semantics or recognizing faces) but the processes and representational formats by which it achieves those computations.”

The Visions of Cognitive Science topic is intended as a place for integrative and reflective papers on cognitive science. More such papers are under development for future issues of topiCS. Those who wish to propose an area or author for a future Visions paper should contact the Editor or an Associate Editor.