- About Us
- Affiliates and Liaisons
This article focuses on the interdisciplinary nature of cognitive science and aims to expose students to one perspective on cognitive science as a field of study.
If you'd like to contribute an article of 1000 words or fewer, please email Izabella Kenny with your article attached.
"The Science of the General Case"
Dr. Robert Goldstone
Associate Professor of Psychological & Brain Sciences
Director of Undergraduate Studies, Cognitive Science Program
Back in the Winter of 1982, Inspired by having just read Douglas Hofstadter's book Gödel, Escher, Bach at the impressionable age of 18, I decided that my undergraduate college's "prefabricated" majors were not what I had in mind, and that I would create my own independent major in the little understood and littler known field called "Cognitive Science."
Thirteen years after receiving my B.A. in Cognitive Science and 21 years after the first Annual Conference of the Cognitive Science Society, I am pleased to see that the field is still going strong. It has even become sufficiently "establishment" that some fortunate undergraduates can become cognitive science majors without having to justify their quirky choice of field of study to three tiers of university curriculum committees, like I did. I am even more pleased to see that my original reason for becoming a cognitive scientist is as relevant today as it was back in 1982.
Cognitive science is the science of intelligent systems. Humans, notwithstanding some notable counterexamples from the United States Congress, provide the best examples of intelligence that we have available. However, for me, the insight that we get from cognitive science is that we may not be the only case of intelligence. Sufficiently powerful computers may some day be intelligent, animals already show signs of intelligence, and it may even be possible to view organizations such as businesses and economies as demonstrating intelligent emergent behaviors. The special case of human intelligence is theoretically important and is also of particular parochial (self-)interest for us, but we should not lose sight of the more general goal of understanding how intelligent systems are possible in any form at all.
Generally speaking, a good heuristic for discovering the nature of a phenomenon is to find as widely different examples of the phenomenon as possible. To find the Universal Grammar shared by all human languages, one would be wise to study Swahili, Chinese, and Dutch, rather than three languages with a common root such as Spanish, Italian, and French. To discover the general nature of orbital systems, one might study both atoms and star systems, because of (not despite) their distinctly noticeable size differences. By the same token, if one is interested in the nature of intelligence, adaptation, or consciousness, one is well advised to study them by using as diverse examples as possible. By doing so, the deep, shared properties of these phenomena are highlighted.
Imagine each example of a phenomenon as a spotlight that includes both critical elements that make it an example of the phenomenon as well as idiosyncratic elements. If the examples are well separated (see illustration below), then the critical element, indicated by the box in the center, will be highlighted as one of the few elements illuminated by all three spotlights. However, if the examples are close, then the critical box may be hard to find amidst the large overlapping area covered by all of the examples. Studying a phenomenon from widely different vantage points allows its essential nature to become selectively emphasized.
Diverse examples of a phenomenon allow us to more efficiently focus on the essential elements of the phenomenon by reducing the(shaded) area that must be searched.
I have heard some of my non-psychologist friends complain at the Annual Conference of the Cognitive Science Society that the meetings have become progressively biased toward psychology. I take these complaints seriously because I believe that Cognitive Science remains a vital field precisely because of its diversity of approaches and opinions. I advocate diversity not on grounds of political correctness, but simply because it's good science.
If a neural network modeler wants to present a model without explaining what ramifications the model has for understanding intelligence or having to justify the assumptions made by the neural network approach to those not in the clique, then let them present their work at one of the many specialized neural network conferences.
I prefer The Annual Conference of the Cognitive Science Society because people there grapple with core issues that span philosophy, psychology, computer science, neuroscience, and linguistics. Researchers there are expected to discuss the assumptions made by their approach and to relate their work to issues more general than a particular engineering solution or experimental paradigm.
If a biologist shows me that the changes over time in grass and rabbit populations are well accounted for by Lotka-Volterra equations (dN/dt = rN-gNP and dP/dt=hNP-mP, where N is the amount of grass, P is the number of rabbits, and r, g, h, and m are parameters governing the growth, death, and energy transfer rates), then I may become interested because of the system's elegance and ability to predict a complex ecological phenomenon.
However, if a doctor shows that the same equations can account for how the number of antigens and antibodies vary over time, or a psychologist shows that the same equations can account for the timing of two behaviors in a pigeon (Bill Timberlake, personal communication, April 1997), then I will really stand up and take notice!
Finding the same equations across domains suggests that they are not simply an idiosyncrasy found in a particular system, but rather reflect the deep nature of systems that have both antagonisms and dependencies between elements.
I am not suggesting that the details of a particular domain are unimportant. I am advocating cross-disciplinary accounts, but these general accounts can only be found by contemplating specific situations. The Lotka-Volterra equations first had to be developed from and informed by ecological data before they could be extended to economics, medicine, and psychology. Genuinely influential research gets the details right within its domain, and also raises questions and answers for other disciplines.
The recent popularity of neuroscience is justified. Neuroscience can offer insights into mechanisms that create minds that could not otherwise be discovered. However, I part company with my neuroscientist friend who said, "Once we understand completely how the nervous system works, then we'll know what minds are." Even granting that this method could produce laws that bridge the neural and the mental, at the end we would only have an understanding of one special case of minds. If we are interested in the general nature of minds as well as the special, human case, then we will have to turn to approaches that allow us to isolate essential from non-essential parts of the nervous system.
The task of winnowing essential from non-essential aspects is not done most efficiently by concentrating exclusively on neural structures. Rather, our selection is accelerated by finding out what other disciplines' candidates for essential elements look like. As the figure above shows, the more distant the disci plines, the better able they are to constrain each others' set of candidates. This is why the cross-disciplinary methods of cognitive science are ideally suited for understanding the essential nature of the mind, in any form that it takes.