Thursday, February 28, 2013

The Metaphysics of Propositional Attitudes

Think of a tiger.  Alright: now, how many stripes does your imaginary tiger have?  Probably your “mental image” of a tiger turned out not to have a specific number of stripes.  Representational theories of mind hold that it is literally true that cognitive states and processes include representations as constituent parts.  To some this may seem self-evident: isn’t remembering one’s mother, for example, a matter of inspecting an image of her “in the mind’s eye”?  Isn’t imagining a tiger similarly a matter of composing one’s own, private image of a tiger?  But there are good reasons for thinking that mental representations must be formal, like linguistic representations, rather than isomorphic, like pictorial representations.  Formal representations, like novels, have the flexibility to include only relevant information (“The Russian guerillas rode down on the French encampment at dawn”), while isomorphic representations, like movies, must include a great deal of information that is irrelevant (How many Russians, through what kind of trees, on horses of what color?).  While there are those who argue for isomorphic representation, most cognitive scientists believe that mental representations must be rule-governed sets of symbols, like grammatical sentences of language.  “Cognitivism,” while a broad term, can fairly be defined as the view that cognition is an internal (neural) kind of information processing using symbols: that is the cognitivist’s model of brain function. 

          Noam Chomsky is perhaps the most influential founder of cognitivism which at the time (late 1950s-early 1960s) was conceived as a challenge to the dominant behaviorism.  Cognitivists like Chomsky argued (as Descartes had 300 years earlier) that language distinguished humans from non-human animals in such a way that, unlike non-linguistic animals, humans could not be modeled as stimulus-response learning mechanisms as (very roughly) the behaviorists maintained.  Chomsky, critiquing B. F. Skinner’s account of “verbal behavior,” made the enormously influential proposal that formal syntactical structure was “generative”: grammatical frameworks like “The_is_the_” allowed for multiple inputs and thus for the generation of indefinitely many (linguistic) representations.  Thus human speech was not “tropic”: the tropic calls of non-linguistic animals were determined by natural history and environmental stimuli, but human speech was, through its generative nature, liberated from these natural determinants.  According to Chomsky, this generative grammar was the cognitive foundation that enabled humans to have mental lives of a qualitatively unique sort.

The early, canonical Chomsky under discussion here held that while the behavior of non-linguistic animals could indeed be explained solely through appeal to behaviorist learning models and biological determinants, the behavior of humans could not.  (In recent years Chomsky has modified this view somewhat, but mostly by acknowledging that some non-linguistic animals have conscious experiences.  Consciousness will be discussed in Chapter Three.  I am not aware of Chomsky anywhere repudiating his original position that non-linguistic animals cannot be said to have intentional states.)  Taken to its extreme this argument appears to show that it is necessary for a being to have an innate (“genetic”) syntactical system for generating propositions in order to be in an intentional state at all (this is what was at stake in the famous primate sign-language research, which was initiated by behaviorists seeking to debunk Chomsky by showing that grammar could be learned by non-linguistic primates).

When Bertrand Russell coined the phrase “propositional attitude” in his 1921 book The Analysis of Mind he wasn’t thinking of “proposition” in the sense of a piece of language.  He was thinking that what was represented was a situation or what would today most likely be called a “state of affairs,” a way the world (or some part of the world) could be.  The latter-day cognitivists took a much more literal view of propositional attitudes as linguistic entities.  Their claim was that the syntactical structure of language was, in fact, the basis of representation in the first place.  On the cognitivist conception of intentional states (beliefs, desires, etc) all intentional states can be described as attitudes towards propositions.  By “attitudes” here one means attitudes towards the truth values of propositions: To believe that the drinking fountain is down the hall is to have the attitude towards the proposition “The drinking fountain is down the hall” that it is true, and to have a desire for water is to have the attitude towards the proposition “I will drink water soon” that one intends to make it true, hopes it to be true and so on.  Propositional attitudes are understood as attitudes towards content that can be expressed in “that”-clauses: one has a belief that “The cookies are in the jar,” a hope that “There is milk in the fridge,” a fear that “Mom will say no to eating cookies,” and so on.

The defender of propositional attitudes argues that intentional states can only be individuated by virtue of their respective contents.  What makes Belief X different from Belief Y is that X is about Paris, say, and Y is about something else.  This looks like a block to reduction: to correlate electrochemical activity in the nervous system with Belief X, we must already be able to specify which belief Belief X is (for example by asking an awake subject what they are thinking while their brain is being simultaneously scanned).  We don’t have any way of getting from no-content to content (from the non-mental to the mental).  This motivates the contemporary version of the problem of mental causation: it appears that the content (the meaning) of the proposition is what plays the causal role in the production of behavior: when told to proceed to the capital of France he went to Paris because he believed that “Paris is the capital of France.”  All the explanation in physical (neurophysiological) terms one could possibly make wouldn’t be explanatory if it didn’t at some point reveal the meaning that is expressed in the proposition, and it doesn’t: “He believes that Paris is the capital of France” is not shorthand for a causal chain of neurophysiological processes.  This is an argument for the ineliminable role of the semantic property: “intentional realism” is the view that mental representations are an ineliminable posit of cognitive science.

Donald Davidson famously pointed out a further problem for the development of “psychophysical laws” (as he called them), laws that systematically identified brain processes with particular contents of intentional thought: no one propositional attitude could ever suffice as the discrete cause of a behavior because the causal implication that the propositional attitude has for the acting subject necessarily emerges from the logical relations that that “attitude” has with all of the other intentional states of the subject (this is the point where the problem of representation and the problem of rationality connect).  Davidson’s phrase for this was “meaning holism,” the view that meaning (inasmuch as meaning can be posited as something playing a causal role in the production of behavior) is a property of networks of propositional attitudes, not of individual ones.  There is not an assortment of individual intentional states that constitute a person’s mental content such that one or another might be identifiable as the sole cause of behavior (although one might be the proximate cause): each person has an intentional economy, if you will, and our immediate reasons for acting are the running product of the unfolding logical relationships among this network of propositional attitudes.

Finally, getting down to ontological bedrock, some philosophers argue that since physical entities and processes don’t have semantic properties (the properties of meaning something, of having truth-values and of having logical relations with each other) then there must be some other sort of entities that do.  They nominate propositions, considered now as matter-independent, mind-independent entities.  This dualist suggestion is usually called “Platonic realism” about propositions.  Propositions, according to this view, are not identical to their corresponding sets of physical tokens (sentences).  The idea is that, as with proofs of mathematics, the entire set of physical tokens of a given proposition (written, spoken or otherwise) could fail to exist without effecting the existence of the transcendental, non-physical entity that is the proposition itself.  These philosophers invite us to consider all of the propositions that have never been uttered or written, or even thought: aren’t they in some sense nonetheless “there,” just as there are certainly many still undiscovered proofs of mathematics?  On this view propositions, like math proofs, are eternal, unchanging, mind- and matter-independent non-physical entities that possess the non-physical (semantic) properties that the propositional attitude model of intentional states (or for that matter any representational theory of mind) entails.

Those of us with an interest in metaphysics are willing to at least consider this sort of suggestion on its own terms.  Hard-headed experimental cognitive scientists, on the other hand, perhaps already impatient with the very idea of metaphysical problems as such, may feel that now this discussion has gone too far: “We’re not Platonic dualists, for heaven’s sake!”  But it’s not obvious that representational cognitivists can really distance themselves from this kind of anti-naturalism (patent dualism in fact).  The representational theorist of mind needs representations precisely because representations, and only representations, are the kinds of things that could possess semantic properties.  “No computation without representation,” as Jerry Fodor put it with his characteristic impishness.

The representational cognitivists, then, appear to be at least tacitly committed to the position that intentional (semantic) properties are real and non-physical, and this position entails further ontological commitments.  Of course anyone is free to embrace some kind of dualist ontology if that is what they are convinced the world is like.  But I, for one, will opt for some kind of monist ontology if that is one of the plausible options.  I would sooner try to eliminate non-physical properties from the theory of mind than grant the existence of the non-physical entities that must exist to have them if it turns out that we can do without the original non-physical properties in the first place.  And while one may pursue the satisfactions of philosophy for their own sake the fact is that these ontological difficulties need to be worked out if psychology (or at least cognitive science) is ever to be naturalized.       

In any event, one can summarize the standard representational cognitivist view as holding that representation is necessary for intentionality (it’s the representations that mean something and this semantic content plays an ineliminable causal role), that syntactical structure (formal rules of composition) are necessary to generate representations and that, therefore, the central project of cognitive science is the investigation of this syntactical structure.  In the late 20th century this cognitivist paradigm supported a great flowering of work in theoretical psychology (much as the behaviorist paradigm had supported a great flowering of experimental psychology in the earlier decades of that century).