PHL471: A Summary of Ontological Positions

A comparison of ontological views

In most of the discussion we are having, and in the papers we are reading, we are typically starting with the assumption that we have a body of very productive and useful scientific theory (that includes things like physics and chemistry and so on), and we're wondering how these strange mental phenomena, for which we have as yet no full explanation, might fit (if at all) with that body of theory.

Thus, I use "physical" here as a gloss. I use "physical" here to stand in for something like "of the kinds that are described in our best (scientific) theories." Also, I'll assume the agents in question are always human. So "x" is some arbitrary human being or human mind.

Theory DescriptionSimplistic Example of an Identity ClaimExample PhilosopherNote
Interactive Substance Dualism Mind and body (extended stuff) are two different kinds of substances that do interact casually. Substances are kinds of beings that can exist without other kinds. For Descartes, there can be minds without bodies, and bodies without minds. (Mental event m is a different type of thing than is any physical event p, and vice versa; thus:)

Mental kind pain for x = some specific kind of mental event in mind of x which is not any physical event and could in principle exist without any physical event.
Descartes Substance dualism without interactivism includes parallelism (the two substances act in ways that, typically through the grace of god, are miraculously seemingly interacting).
Philosophical Behaviorism Mental state terms and descriptions are translatable without significant loss of utility into some behavioral state terms or descriptions. (One might say, minds are behaviors, but some philosophical behaviorists believe we should avoid talking as if there is a thing, mind.) pain for x = the increased likelihood of x to withdraw, to say "ouch," to undergo certain autonomic changes, etc.

Applied to belief: it's hard to say something general about belief accounts as a whole, but a specific case might allow an illustration of the idea applied to belief. Thus:

belief by x that there is food in the refrigerator = the increased likelihood of x to open the refrigerator when looking for food or when hungry, to say "the refrigerator" when asked "where's the food?", etc.
Ryle Philosophical behaviorism is distinct from psychological behaviorism both by the linkage with a focus on language, and also a presupposition that (almost) all our mental state talk can be saved and translated into behavioral talk.
Psychological Behaviorism Some of what we call mental events are predispositions to kinds of behavior in response to kinds of stimuli; some of these dispositions are learned through conditioning. Or: some mental terms and descriptions are translatable to some behavioral event terms or descriptions. pain for x = the increased likelihood of x to withdraw, to say "ouch," to undergo certain autonomic changes, etc. Watson, Skinner The feature that distinguishes psychological from philosophical behaviorism is both an openness to the idea that we might introduce new mental or behavioral terms; or revise some of those that we have; and that we may eliminate some of our mental terms as unsalveagable; and an occasional willingness (in practice, but not in their descriptions of their own method) to allow for some internal mental events or capabilities. Also, as a practical matter, psychologists were concerned with method and not ontology.
Type Identity Theory Some mental kinds are physical kinds mental event kind pain = physical event kind of such and such neurons firing. (Here, the physical event is left even more vague because we don't have yet much idea about how to do such a reduction.) Smart, Place Types may be of varying specificity. Some might for example think that there are physical types across species for a mental type like pain (e.g., all animal pains are the same kind of mental event, and identical to the same kind of brain event). Others might construe the kind more narrowly; e.g., supposing that a kind like pain is specific to, say, a species, so that human pain would be identical only to a physical kind found in humans.
Proper Token Identity Theory Some mental events are physical events (but for some of these mental events there is no kind that is also a physical kind) mental event instance pain of x = physical event instance such and such neurons firing in x. (Here again I made up and left vague a physical event. The important thing is that each instance is reducible, but -- in the case of what I'm calling "proper" token identity theory -- the kind is not.) Davidson Note that type identity theory would entail token identity; so, when a theory is called a "token identity theory" it is typically assumed that it is not a type identity theory. Token identity theory typically depends upon some notion of multiple realizability.
Functionalism Some mental events are physical events, which can be identified by the functional role they play in a functional system. pain of organism x = physical event (probably a brain event) that plays such and such a functional role in leading the organism to be more likely to withdraw, to say "ouch," to have certain autonomic changes, etc.

(We would ultimately describe the functional role in terms of the relation of this physical event to other physical events that have also their own functional roles; functions are products of a network of such relations.)

belief that P for organism x = physical event that plays such and such a functional role in leading the organism to be more likely to assent to P, etc., and having relationships to other functional events of a kind that reflect the semantic relationships between P and those events.

This last example is more of a punt, but the idea is that the relation between P and other propositions would be reflected in the functional organization. For example, if you believe that P-->Q and you believe that P you are inclined to believe Q, and so on.
The important difference from behaviorism, which must be stressed, is that there is an event which is causing the relevant behavior, which has the relevant relationships. In behaviorism, the behaviors stand alone and we posit no such event. It is debatable, and depends on details, whether functionalism is a token or type identity theory. The functional events referred to are types, but whether those functional types need to be physical types is debated (I think they must be, but not all agree here). The primary inspiration for this view is the computer; software events are excellent examples of functional events.
Eliminativism Some mental state terms do not properly refer to anything. They are superstitions, like ascribing an illness to demon possession. [For example, assuming eliminativism about belief] There is no thing x such that x = belief that P. Rorty and Churchland on some propositional attitudes, some psychological behaviorists for some mental state terms Restriction of the domain to be eliminated is carefully done by eliminativists. It would be a parody to say they eliminate the mind.

A brief review of the logic of identity.

It is possible to define identity of particular objects using second order logic. This is sometimes called "Leibniz's law":

∀x∀y ( x = y ↔ ∀F ( Fx ↔ Fy ) )
(This is called "second order" because "∀F" means for any property. Normally the "∀" just means for any object.)

If you find it difficult to recall your logic, think of it this formula as saying: "x and y are the same if and only if for any property F, x is F just in case y is F."

You should know Leibniz's Law: it keeps us honest by reminding us that two particulars are the same just in case they have all the same properties. Identity claims are very strong.

Two additional thoughts:

It is difficult to see how we would express type identity using these tools, but here is a try:
∀F∀G ( F = G ↔ ∀x∀H ( ( Fx ↔ Hx ) ↔ ( Gx ↔ Hx ) )
It's hard to say that in English; I'm not going to try. Also, I'm frankly not sure that's air tight. I'll look and see if another logician has tried it.

Finally, it is challenging to see how to express in the rigor of logic Davidson's notion of token identity without type identity. Right now I see no way to express it that is not consistent with functionalism.