1
Standard cognitive science
- 1.1 Introduction
- 1.2 Newell and Simon’s General Problem Solver
- 1.3 Descriptive frameworks
- 1.4 Back to General Problem Solver
- 1.5 Sternberg’s analysis of memory scanning
- 1.6 The computational vision program
- 1.7 The solipsistic view
- 1.8 Summary
1.1 Introduction
What I shall be calling standard cognitive science has roots in a number of disciplines, including computer science, psychology, linguistics, and philosophy. Because my interest is less in tracing the history and development of standard cognitive science than in providing an idea of its theoretical and methodological commitments,1 I propose to introduce some important ideas within standard cognitive science through discussion of several exemplary research projects. These projects are (i) Allen Newell and Herbert Simon’s General Problem Solver; (ii) Saul Sternberg’s work on memory recall; and (iii) computational analyses of perception. Despite the diverging explanatory targets of these three enterprises, they are remarkably similar in how they conceive of the process of cognition and in their commitments to how cognition should be studied.2
1.2 Newell and Simon’s General Problem Solver
In the early 1960s, Newell and Simon created a computer program they called General Problem Solver (GPS),3 the purpose of which was not only to solve logic problems, but to solve them in the same way that a human being would (1961, 1976). That is, the program was intended to replicate the internal thought processes that a human being undertakes when solving a logic problem. As such, just as Newell and Simon claim, GPS is a theory of human thinking (1961: 2016). Examination of some of the details of GPS thus provides us with a picture of what the mind looks like, for a cognitive scientist.
Because the purpose of GPS was to replicate the stages involved in human problem-solving abilities, its assessment required that the problem-solving procedure it used be tested against the problem-solving procedures that human beings use. Thus, Newell and Simon asked human subjects to “think out loud” while solving logic problems. For instance, a subject was shown a logical expression, such as
and was asked to transform this expression into
The subject was also provided with various transformation rules of the sort that would be familiar to anyone with a background in simple sentential logic. No interpretation was offered for the logical expressions. Subjects merely had to identify the rules that would transform one syntactical object into another and then apply them.
Obviously, transformations of expressions like (1) into others like (2) are a simple matter for any suitably programmed general purpose computer. As Newell and Simon describe, a computer
is a symbol-manipulating device, and the symbols it manipulates may represent numbers, words, or even nonnumerical, nonverbal patterns. The computer has quite general capacities for reading symbols or patterns presented by input devices, storing symbols in memory, copying symbols from one memory location to another, erasing symbols, comparing symbols for identity, detecting specific differences between their patterns, and behaving in a manner conditional on the results of its processes.
(1961: 2012)
The hypothesis that motivates GPS as a theory of human thinking is that thought processes are just like those processes that take place within a computer:
We can postulate that the processes going on inside the subject’s skin – involving sensory organs, neural tissue, and muscular movements controlled by the neural signals – are also symbol-manipulating processes; that is, patterns in various encodings can be detected, recorded, transmitted, stored, copied, and so on, by the mechanisms of this system.
(1961: 2012)
Finally, given that human thought processes are computational, that is, that they involve the manipulation of symbols, GPS provides a model of human cognition just in case the kinds of computations it uses to solve a problem are similar to the computations that take place in a human brain, where these latter computations become visible through Newell and Simon’s “thinking out loud” experimental protocol.
1.3 Descriptive frameworks
Before proceeding further with discussion of GPS, it is worth pausing to consider Newell and Simon’s claim that there are symbols and symbolic operations within the brain. In one sense, this must seem obviously false. If one could peer into a brain as it is solving a logic problem, nowhere would one see letters and connectives, or symbols of any sort. What, then, does it mean to say that human thought is symbolic, or that the processes that take place in a human brain are analogous to those that occur in a computer?
One quick way to defuse this worry is to point out that as hard as it is to find symbols in the brain, finding them in a computer is no easier. Open up a computer and you are no more likely to find letters, numerals, and other symbols than you are when you dissect a brain. But, of course, computers are symbol processors par excellence. So, where are the symbols?
The way out of this mystery requires the recognition that the world and its contents lend themselves to different kinds of descriptions which, despite their differences, may nevertheless be true and consistent with each other. For instance, the coffee mug sitting dangerously close to my keyboard is a collection of molecules. But it is also a mass of ceramic. And, of course, it is also a coffee mug. It is also my coffee mug. All of these descriptions are true of the object on my desk, and the truth of one description does not exclude the truth of the others. Precisely why the various descriptions of my mug “fit,” why they are each true of it, turns out to be a difficult issue. Part of the explanation will involve chemical facts, and another part, facts from materials science, while still other parts of the explanation will depend on facts about, perhaps, the individuation of artifacts and the social contracts that justify certain property rights.
Although obvious on reflection, the idea that items in the world can be described in distinct but compatible ways opens the door to the possibility that neurons, or their activities, can be symbolic. Whether they are symbolic, that is, whether the description of them as being symbolic is true, depends on whether they meet the criteria for something’s being a symbol. The question “What makes a part of the brain a symbol?” is, in effect, like the question “What makes this mass of ceramic a coffee mug?” Some masses of ceramic are not coffee mugs (e.g., the ones to which power lines are attached), but some are. And, while perhaps there may be no certain answer to the coffee mug question in all cases, often enough we have sufficient grasp of the meaning of “coffee mug” to say of a mass of ceramic that it is or is not a coffee mug. If the ceramic is shaped cylindrically, is closed on the bottom and open on top, is impermeable, has a handle, is within a certain range of sizes, then this is usually a good enough reason to describe the ceramic object as a coffee mug.
Now consider the sorts of things that are properly described as symbols. Words are symbols, as are numerals. But why are they symbols? The answer to this question has little to do with the physical material out of which words and numerals are constructed. Words and numerals can be scratches of graphite on paper, illuminated pixels on a computer monitor, or light bulbs on a billboard. The conditions for being a symbol do not, apparently, mandate that symbols be made of any particular stuff, but instead focus on functional criteria, on criteria about what symbols must do. Among the relevant conditions seem to be these. First, symbols can “stand for” or represent things. The words “lamb” and “αρνι´” are both symbols, and both happen to represent the same wooly object. Similarly, the numeral “12” represents the number twelve, as does “XII.” And while “X” can represent the number ten, when found on a map it can represent the location of a treasure chest. A band of gold, when worn on a finger, can also be a symbol, representing, as it often does, devotion.
In addition to having a representational function, sometimes symbols can be combined with others in accordance with various rules, such that new symbols result.4 The words “Mary,” “had,” “a,” “little,” and “lamb” can be combined, according to rules of English grammar, to form the sentence “Mary had a little lamb,” which represents the state of affairs of Mary having a little lamb. The rules that permit the combination of those symbols to form that sentence also prohibit the combination that results in “Mary a lamb little had.” Similarly, rules of arithmetic permit the construction “12 + 45 = 57” but do not allow “= 12 45 + 57.” These obvious points suffice to illustrate a feature of some symbols: they may be combined according to rules that dictate which operations upon them are “legal.”5
The fact that some symbols display combinatorial properties is especially important for understanding standard cognitive science, for the idea that thinking involves operations on words in a “language of thought” (Fodor 1975) has become mainstream. In this view, thoughts are sentences in an internal language, and reasoning involves combining and manipulating the components of these sentences, just as one might do when performing syllogisms in a natural language. As we shall see in the next chapter, some critics of standard cognitive science have sought to undermine the idea that cognition relies on linguistic structures.
The combinatorial power of symbols, as well as their capacity to represent or stand for objects apart from themselves, are, of course, not intrinsic features of the symbols on their own. Scratches of graphite on paper or sequences of on- and off-currents in a computer gain these capacities in virtue of some sort of controller – a human agent, perhaps, or, in a computer, a central processing unit (CPU). A series of on- and off-currents becomes a string of symbols because the CPU treats it as such – combining and transforming members of the series according to a rule-prescribed grammar, which in turn may be encoded in electrical current that the CPU has been designed to read.6
A final point about symbols, and one that will loom larger in the context of some theories of embodied cognition, is that their relation to their content – to what they represent – is often arbitrary. The word “lamb” represents lamb, but not because the word and the animal are similar. The word is neither white, nor fluffy, nor best when grilled on a spit and served with lemon. Likewise, the particular assignments of numerals to numbers that is present today was by no means inevitable (Romans represented numbers with different symbols than those we use). The numeral “6” might as well have been used to represent five as it was to represent six. Of course, this point about the arbitrary connection between symbols and their contents raises a hard question about how one thing can come to be about something it does not resemble, but we needn’t wade into those troubled waters for the purposes at hand.7
We have available now enough pieces to make sense of Newell and Simon’s speculation that the human brain is, like a computer, a symbol- manipulating device. The elements of the brain are, under one description, neurons, but, if they behave in a certain way, they might also be symbols. And, Newell and Simon suppose, (some) neurons do behave in such a way that a description of them as symbols is true. Neurons, singly or in a group, represent – they stand for things in the world in virtue of processes that begin with the stimulation of the body’s sensory organs. Presumably neurons that code for one concept can be combined with neurons ...