In 'Troubles with Functionalism', Block introduces functionalism as a "new incarnation of behaviorism" where "sensory inputs and mental states" produce a "disposition to act and have certain mental states" (262). The functionalist seeks an ontological reduction of mental states to functional states viz. functional descriptions of their logical relations grounded1 in physical (sensory) inputs and (e.g. behavioral) outputs. Such functional states could be modeled in terms of logical sentences like Ramsey functional correlates, or as parts of dynamic (state) models or computational systems like Turing machines. An actual machine (or nonmachine) which implements such states, sensory inputs and behaviors (outputs) can be said to realize2 the corresponding model.
Block distinguishes two methods of reduction: Functionalism, where inputs, outputs and states correspond to common-sense or folk-psychological theory, and Psychofunctionalism where inputs, outputs and states correspond to scientific psychological theory (allowing e.g. neural activations as an input or output) (268-9). The quote on p. 292 occurs in a context where Block has just contrasted two types of homunculi: one composed of subatomic aliens scientifically indistinguishable from an ordinary human (realizing a Psychofunctional model of brain activity), and one whose inner workings (perhaps fleas, tiny men, or a Turing machine) are empirically completely unlike an ordinary human (Blocky, realizing a Functional model of human behavior). Here we can regard the Functional simulation as one that fulfills a reduction of folk-psychological states and behaviors, and the Psychofunctional simulation as one that fulfills a reduction of neurological states and interactions.
Troubles with Blocky
Block implies that a Functionalist should accept ex hypothesi that Blocky has Functional states and therefore (folk-psychological) mental states identical to a human. However, there is room for disagreement. As a simple example, if Blocky were constructed to simulate person A, A may know themselves to have a genuine psychology including a genuine quale for pain. However, they may have reasonable doubt as to whether Blocky has the same quale or even any qualia at all. If we follow a Kripkean notion of identity we can then trivially conclude that it's not necessary that Blocky has the same mental state.
In another instance, Block gives an example of an intermittent paralytic on page 297. Consider a different (not Block's) intermittent paralytic Mitty with a neurological condition which often prevents him from drinking water or realizing that he is incapable of doing so. Mitty in one instance desires to drink a glass of water and believes himself capable of doing so. But a paradigmatic Functional interpretation of believing oneself capable of drinking water and desiring to drink water may be to typically result in drinking water. However, if Mitty typically fails to drink water, a corresponding Blocky under that paradigmatic Functional interpretation would not be ascribed the functional state of both believing oneself capable of drinking water and desiring to drink water. Arguably, this corresponds to a difference in information-processing between Mitty and Blocky3.
Shoemaker suggests for the case of paralytics etc., considering a "paradigmatically embodied person," which Block claims to be a retreat from Functionalism to Psychofunctionalism (298-9). It seems we can infer that Block would say that in terms of information-processing, a Functional Blocky is not necessarily the same as the corresponding human (Mitty). The case is different for information processing in the context of a Psychofunctional homunculus, not based on folk-psychology. There we might exchange the terms of belief and desire for empirical expressions in an appropriate scientific psychological theory. Following Block, it seems plausible to accept ex hypothesi that such a simulation would necessarily have the same empirical psychology (information-processing) as a human.
Mental Differences and the Argument against Functionalism
The quote from p. 292 seems to suggest a coherent argument against Functionalism. In the first part, things like Blocky "are not things to which neurophysiological theories true of us apply." Most relevant here is that if humans are A, and Blocky B, that from a neurophysiological perspective A and B are completely different. So, if you think that neurophysiology is relevant to mental properties, you should expect that from a mental perspective A does not equal B.
"They need not be things to which psychological (information-processing) theories true of us apply," appears to be a modal claim. We can express this as saying that from an information-processing perspective it is not necessary that B is equivalent to A.
Here, we can simplify and formalize the argument to show how one would argue for mental differences and how to use those differences to argue against Functionalism. For convenience, we'll refer to the following contexts: neurophysiologically (N), information-processing (I), qualia (Q), mentally (M), and functionally (F). We'll also include a human A, and Blocky B and use modal logic operators
<> for necessary and possible respectively.
Given a _Functional_ Blocky, we obtain some simplified statements from the quote:
1. (N) A =/= B
2. (I) ~ (B = A)
And transform them a bit
2a. (I) <> ~(A = B)
Since properties of A are fixed
2b. <> Exists B s.t. (I) ~(A = B)
2c. <> Exists B s.t. (I) (A =/= B)
But if we take such a B to be theoretically plausible, and as the B under consideration then
2d. (I) (A =/= B)
So we have
3. (N, I) (A =/= B)
Meaning that neurophysiologically, and from an information-processing standpoint, humans are not equivalent to Blocky. Block elsewhere supports the intuition that there's a corresponding absence of or difference in qualia.
*3a. (Q) (A =/= B) or (B has no qualia)
Although it doesn't necessarily follow, Block thinks it's reasonable that we have intuitions that Q, N and I are relevant to mental state. If so, perhaps we can conclude:
*3b. (M) (A =/= B)
Alternatively, *3a as (B has no qualia) can satisfy the premise AQ of the Absent Qualia Argument (AQA)
From _intuition_, _prima facie_
AQ. (<> Exists B, B has no qualia)
But since (presumably) mentality requires qualia
*3b. (M) (A =/= B)
With *3b we can make the argument against Functionalism:
We introduced Blocky as Functionally equivalent to A
4. (F) A = B
But if we want to use functionalism as a reductive theory of mind, then we'd expect mental equivalence to be implied by functional equivalence. We'd say that
*5. for all X, Y (F) X = Y -> (M) X = Y **or** for all X, Y (F) X = Y <-> (M) X = Y
so for A and B we can say
5b. (F) A = B -> (M) A = B
But if we believe *3b, then
6. (F) A = B and (M) A =/= B
So if *3b is reasonable, we should deny *5.
Block's distinction between Functional and Psychofunctional simulations alongside his phrasing of "psychological (information-processing) theories" appears to emphasize e.g. neuroscientific theories of how we process information. This seems to correspond to how Blocky's internal logic is not the same as the logic realized inside a human brain.
In response, a functionalist (perhaps Psychofunctionalist) like Shoemaker would accept the AQA but deny *3a and so deny *3b. An eliminativist could accept AQ and *3a but deny that *3b follows. Block seems to think AQ, *3a and *3b are all reasonable.
1 See symbol grounding problem. Block describes this as "'tacking down' mental states only at the periphery." (284)
2 Realization is a standard term for an implementation of a computational model (e.g. a Turing machine) or other input/output system in mathematical or computational logic. See also Realization (systems), and p. 306 for an example for Block's use regarding nonmachines.
3 To expand a bit on this, one common folk-psychological theory that seems suitable for a Functional law relates belief, desire and action. We might say the desire to φ along with the belief that one can φ, typically results in doing φ. So a paradigmatic Functional interpretation of belief and desire would be belief that can φ + desire to φ -> typically does φ. We'll call this Functional law belief+desire->action. Since Blocky is a simulation of Mitty, Blocky typically fails to drink water. Yet since Blocky is a Functional simulation he follows Functional laws and has corresponding mental states. But because of belief+desire->action, Blocky to be a Functional simulation must lack either the desire to drink water, or the belief that he is capable of drinking water. If Mitty doesn't drink because he can't while Blocky doesn't drink because he either doesn't desire to or doesn't believe that he can, this seems to correspond to a difference in information-processing between Mitty and Blocky.
This further seems to suggest a difference in mental states between Mitty and Blocky. A Functionalist could respond that Mitty doesn't actually have belief+desire for drinking water. But that seems to disagree with common-sense psychology. Alternatively, one could deny belief+desire->action. But we are left with a similar problem for any mental state that can't be expressed in a paradigmatic Functional way.