#architecture mind

[maaarine]

The Architecture of Mind (Peter Carruthers, 2006)

“Virtually all of the scientists who study human reasoning and the pervasive fallacies that so often occur in human reasoning have converged on some or other version of a two-systems theory.

System 1 is really a collection of systems, arranged in parallel.

These systems are supposed, for the most part, to be universal (common to all members of the species), to be evolutionarily ancient, and to operate swiftly and unconsciously.

Moreover, their processing algorithms are either immutable, or subject to their own idiosyncratic trajectory of learning and change—at any rate, explicit instruction has little impact on their operations.

In the context of the present discussion, then, they can be identified with the set of central / conceptual modules. 

System 2, in contrast, is supposed to operate linearly (rather than in parallel), and to be slower and characteristically conscious in its operations.

But it can override or pre-empt the results of System 1.

And its algorithms are much more mutable, and are more easily influenced by explicit teaching of various sorts.

System 2 is also much more subject to individual variation.

Thus Stanovich (1999) shows that variability in success in the various standard reasoning tasks (which are thought to require the operation of System 2) correlates highly with IQ, and hence also with g.

And even when IQ is factored out, it correlates highly with certain measures of variable cognitive style (such as a disposition to be reflective, and a capacity to distance oneself from one’s own intuitive views)."

[maaarine]

The Architecture of the Mind (Peter Carruthers, 2006)

“One important fact that should be emphasized in this context is that autistic children can pretend when prompted to do so, but they appear to lack any motivation for, or interest in, pretending—they say that they can’t see the point.

Admittedly, autistic children are also somewhat less good at pretending than are normal children. (…)

Most developmental psychologists believe that mind-reading emerges in stages.

Infants have the capacity to identify intentional agents, and to attribute simple goals to such agents.

Two-year-old children understand some aspects of perception and desire, but still lack the capacity to solve false-belief tasks.

Then four-year-old children have a fully representational understanding of mind—they know that beliefs are representational states that can partially represent a fact or situation, or that can misrepresent it. (…)

What might be predicted concerning the pretend behavior of autistic children? Well first, we would predict that they should retain the capacity for pretence.

For there is nothing in our account of the latter that seems to require mind-reading abilities.

But we would also predict, I think, that autistic children might find it difficult to derive the normal motivational rewards from pretend activity, and that they might therefore not engage in such activities much or at all.

For recall why the child from our earlier example finds pretending to be a cowboy rewarding.

By representing the movements that he is making and the sounds that he is uttering as those of a cowboy riding a horse, he is representing himself to be something that he finds admirable; and this is emotionally rewarding.

But if autistic children have difficulty in understanding and representing actions and agency, then they might have difficulty representing what they are doing in such a way; and hence they would derive no emotional reward from it.

Let me work through this example in just a little more detail by way of emphasizing the point.

An autistic boy might well know quite a bit about cowboys and what they do; and he, like many little boys, might find the idea of being a cowboy attractive.

If prompted to suppose that he is a cowboy, then, he should have no difficulty in assembling an appropriate motor schema and acting upon it.

So he skips along with the intention of being a cowboy riding a horse (conditional upon the assumption that he is a cowboy, of course).

So far, there are no problems on the output / behavioral side.

But in order to derive pleasure from what he is doing, the representation of himself as a cowboy has to be received as input by the motivational systems.

And this requires the mind-reading system to be capable of ‘seeing’ his own movements in that light.

More strictly: the boy’s mind-reading system, receiving visual, auditory, and proprioceptive input concerning his own movements and sounds, has to be able to interpret them as the movements and sounds of a cowboy riding a horse.

(This is actually a kind of ‘seeing as’ not unlike seeing a cloud as a dragon, say.)”

[maaarine]

The Architecture of the Mind (Peter Carruthers, 2006)

“Once a set of norms has been acquired, and stored in the norms database, then a set of inferential mechanisms has to be alert for circumstances in which a given norm might apply.

If I have stored the norm, ‘I mustn’t hurt someone weaker than myself’, then I shall have to be on the lookout for circumstances in which this might happen.

And plainly this requires the normative module to interact with other systems, such as one that might estimate people’s relative strengths.

But if norms are stored in the form, not of explicit rules, but rather of paradigm cases or exemplars (as Sripada and Stich, 2006, indicate that they might be), then the reasoning involved will have to be of a rather different kind.

I shall have to be on the lookout for circumstances that are relevantly similar to one of my stored moral paradigms, and then figure out what action would be required of me if I am to act in the sort of way exemplified.

Most importantly, though, what the normative module does is attach intrinsic motivation to the rules stored in its database.

So when the inferential system accesses a stored norm, and figures out that in order to comply with that norm I must perform some action A, it produces in me in consequence an intrinsic desire to A.

(Of course, as already noted above, this desire has to compete with others for the control of my behavior. So in any particular case I might or might not perform the action A.)

And it does this whether or not I also have extrinsic reasons to A—such as fear of ostracism.

But not only does the system produce an intrinsic desire that I should A, it also produces an emotion of guilt in case I do not.

And it produces anger or indignation if I learn of someone else in such circumstances who fails to A, as well as if I learn of someone who fails to be angry or indignant at such a failure.”

[maaarine]

The Architecture of the Mind (Peter Carruthers, 2006)

“A child sees a banana, activating the mental representation, banana. 

But the banana is somewhat similar in shape to a telephone handset; so the representation, telephone, is also (albeit more weakly) activated.

So far, we might suppose, this doesn’t introduce anything different from what might happen in the mind of an ape.

But now the crucial difference is that each of these representations might be used to activate related action schemata—peeling the banana and eating it; making a telephone call to Grandma.

And each of these schemata is mentally rehearsed. But the child isn’t hungry, so the imagined eating of the banana generates no emotional rewards. 

Talking to Grandma on the telephone, on the other hand, is just what the child likes to do; and imagining doing it is emotionally rewarding. (…)

While the ape’s suppositions (imagined actions) are grounded in its beliefs about current circumstances,

the child’s mental rehearsal of the action schema of making a telephone call to Grandma, in contrast, isn’t grounded in an initial belief that the banana in question really is a telephone.

The mechanism that artificially boosts the weakly activated representation, telephone (or that uses that representation to activate action schemata that are suggested by the affordances of telephones),

must also be one that tags all subsequent representations as mere suppositions, not to be believed outright. (…)

One way of modeling this would be to assume that concepts like telephone and banana are stored along with a set of semantic primitives of various sorts, some of which (such as curved) are shared by more than one such concept. 

Then when banana is activated, these semantic primitives will also be activated, leading to weak or partial activation of any other concepts (like telephone) with which they are connected.

The disposition that underlies pretend play, then, would consist in a willingness to activate and rehearse some of the action schemata that are in turn primed by, or that are affordances of, the resulting weakly activated concepts.”