Please enable JavaScript.

Coggle requires JavaScript to display documents.

Mind-reading (Simulation theory (Gordon- radical simulationism (Gordon'…

- - - - An advantage of 'hot' methodologies is that they are able to deal with ineffable factors like a facial expression very hard to describe- which might not translate so well to a predicate that can be plugged into a 'cold' theory of mind.... the key point is that theory-theory requires categorisation for something to be put into the mind-reading calculus, whereas hot methodology does not
      - 'Hot' methodology could greatly reduce the need for us to have concepts of mind and to recognise concepts in others- we only need to pick up things by contagion of some sort, and use these second-hand belief and desires
    - - Gordon argues against the idea that simulation requires the ability to recognise one's own mental states and then make inferences from our mental states to those of others. Goldman says that we recognise our own mental states by their qualitative 'feel'. Gordon offers an alternative: ascent routines.
        Ascent routines involve using the answer to a question at a lower semantic 'level' to answer the higher question. Gordon gives the example of Mickey Mouse- we are asked whether we believe that he has a tail. In order to answer this question about our belief, we ordinarily ask ourselves simply whether Mickey has a tail- in answering that question, we come to a view about our belief.
        
        Gordon calls the question-within-a-question here an 'object question'- a question about an object which we use to answer a question about ourselves and our mental states
        
        Children can run an ascent routine even if they do not fully understand the mental concept of belief, and so correctly answer questions about belief even if they can't make "genuine, comprehending" ascriptions of belief
        
        Ascent routines provide an intriguing alternative way of answering questions about our mental states without needing to recognise them by qualitative feel or actually getting the relevant 'concept' fully, even
        
        Gordon argues that a similar ascent routine can run for pain- the child understands the pain as a property of his foot and answers a question about the foot, not about his mental states as such. He can answer though not understand questions about his pain through the use of the ascent routine, without need for the concept of pain.
        
        But- if ascent routines don't involve the ability to ascribe beliefs, pains etc. as such, then how do we do this? Gordon suggests the answer lies in the idea of a "mental location". This is the component that children may lack when answering questions correctly through ascent routines. When an ascent routine is used as part of a simulation, this idea of mental location comes in, as the simulator shifts identity to take up the role of the other- thus allowing a reconceptualisation of the object-question in terms of a mental location.
  - - - Argument for simulationism to be used, contra theory-theory
        
        Heal discusses the problem of relevance in the context of the Frame Problem in artificial intelligence- here there is a difficulty getting a machine to give the relevant answer to a question, rather than something true but irrelevant that it has derived from its inference rules plus its body of information- and the larger the body of information, the greater that problem. How do we get a device to ignore true but irrelevant things in particular cases without closing off the possibility of the device referring to those same truths when they are relevant?
        
        This is especially relevant if we concede epistemic holism, as Heal thinks we should do, because then relevance is context dependent and not fixed by rigid rules. Henry 7's Tudor ancestry can be relevant to a measles diagnosis provided there are appropriate linking facts between them that makes it relevant. Anything can be relevant to any decision, in principle.
        
        So, how are the Frame Problem and theory-theory connected? Well, it's not that theory-theory needs to solve the Frame Problem, because that is to a large extent a problem of implementation. But- the theory-theorist does need to suppose that we, humans, contain within OUR theory a systematic way of determining relevance of facts to decisions. And this needs to be a theory, not just know-how.
        
        So what's the problem with supposing that we do have such a theory? One, the oddness of saying that we have it tacitly but have no idea how to bring it to mind explicitly for use in solving the Frame Problem. Two, the truly amazing complexity that such a theory would have. Consider again the massive amount of information in one person's worldview, how finely delineated things may be, and how many people there are. This is a mammoth task, and a mammoth theory if it exists.
        
        The theory-theorist might shy away from this and say that it is not a theory of relevance in that sense that we use- it is our own amazing machinery which sorts through things. But this is precisely to concede the point to the simulationist. We apply our own machinery to the worldview of another.
      - Argument re the limits of simulationism
        
        We return to the distinction between content and non-content. Heal draws the attention to two aspects of mental states which are not representational content: propositional attitudes, and qualitative characters of mental states.
        
        (propositional attitudes are non-content because the representation remains the same- it's our attitude which changes)
        
        Non-content has a 'cluster' of properties- e.g. craving has urgency, it results in action to get the craved item, it's not based on a rational appraisal of the item's value, it can feel unpleasant. These properties, Heal says, are a posteriori in two important senses: one, possession of one feature in a cluster doesn't entail the others, and two, we don't know a priori that any given subject has a particular cluster. The features of a particular subject's psychology are not given a priori in the very notion of a psychology.
        
        Now- for simulation to be successful with non-content, it would have to be possible for it to somehow gain new information about it through some sort of simulation process- because the information is a posteriori, not a priori- we don't already have it merely by possessing a psychology ourselves! The only obvious candidate for an ability to do this would be imagination- we use imagination to simulate a non-content mental state and so gain new information about it.
        
        We have reason, on the 'vivid imagining' view of imagination, to think that this would be unreliable- imagining something is often not exactly the same as the real thing- example of someone who doesn't know about after images in visual experience being asked to imagine seeing something bright- they'd likely not imagine the after image! Because these things are a posteriori and the simulator is not in possession of the real mental state X from which the X-cluster of properties flows, simulation will not be helpful in providing new information.
        
        I don't understand the objection to the other way of conceiving of imagining here- more like a list of properties..? must re read
        
        So: given all below, we need a theory to deal with non-content, but this does not challenge our use of simulation with respect to content
      - Technically Heal's paper seems to be advocating for a hybrid approach
      - Heal's understanding of key terms in the debate:
        
        Theory- set of claims and inference rules, needs to be publically expressible in principle though may be known tacitly, needs to be more than just a set of individual judgements- must be some generalisations in there
        
        Furthermore, Heal takes the relevant concept of theory to theory-theory to be one where parts of the theory are appropriatley connected to each other- wise woman's generalisations vs modern medical theory example
        
        Simulationism- person A can run a simulation of B that is sufficiently similar to be able to infer things about B's mental states and future behaviour
        
        'Content' of a mental state- that representational aspect of a mental state which specifies how the world is or can be assessed for fit with the world
        
        Heal doesn't think that content is all there is to mental states
        
        Heal thinks it fairly obvious that we do possess theoretical knowledge about other people. She just thinks that's not all of what we draw on when predicting and understanding others. Her reasoning for this is that theory alone cannot fully grapple with content. Our theories deal with broad generalisations, not particular subject matters (e.g. medicine, ordering at a restuarant). We may need something else to take us to conclusions about decisions in particular subject matters.
- - - - Also, how could children even operate as little scientists without the concept of false belief? (I don't quite understand this point)
    - - Right- first- how will theory-theory deal with the attribution of mental states to oneself? Well, firstly, it does not need to say that it happens through conscious inferences based on our own behaviour, and this is a good thing because that is highly implausible. Carruthers claims that the idea that less conscious inferential reasoning is going on is plausible, however. Theory-theory says that we have 'immediate' access to occurent mental states which actually proceeds from implicit, unconscious 'reasoning' of some sort.
        
        But- Carruthers says that this can only apply to occurent mental states- things which happen at a particular time, rather than say, long-standing beliefs, intentions or plans. Why? (well, ok I'm not sure)
        
        How do we get knowledge of long-standing mental states then? By turning them into occurent mental states! For example the belief that p is likely to emerge in an occurent judgment that p, of which we can have quasi-perceptual knowledge.
        (this has a flavour of Gordon's ascent routines- but Carruthers seems to want to cash it out in a theory-theory way)
        
        This feature helps us to deal with mis-attributions of belief, where our judgement gets messed with in a way that leads us to get it wrong about our own beliefs. For example, placing effects, where the subjects add a justification post hoc which is at variance with the facts.
        
        There is a paragraph here dealing with one further problem with this theory and a solution to it that I need to understand later
      - Carruthers on Gordon
        
        Carruthers first summarises Gordon's account. He then says that he thinks that one supplementation is needed for Gordon's theory- because pretending that something is the case, representing it so, is not the same as asserting it at all- so how do we get from representing/ simulating X thinking P to asserting that X does think P? Well, by positing an innate disposition to complete simulation by 'transforming pronouns' so that 'I believe P' in the simulation becomes 'X believes P'. This substitution of pronouns works because of sufficient similarity between practical reasoning systems. True, it does involve accepting some kind of tacit inference, but not the problematic introspective kind that Gordon wanted to reject- so Carruthers thinks that he should concede the point.
        
        Now- for Carruthers' objection to Gordon- it lies in the fact that he thinks that Gordon cannot make sense of the phenomenon of our having non-inferential (in Carruthers' sense, not completely non-inferential) access to occurent mental states.
        
        Why? well, Gordon has eliminated introspection entirely in favour of a story about ascending from behaviour (assertions) to self-ascription of mental states- but then, how to make sense of introspection before behaviour? Gordon's theory then becomes objectionably behaviourist.
      - Carruthers believes that the empirical data supports theory-theory- see his conclusion