PMS-WIPS 012 - Robert Thompson - Believe it, or Not? Explaining why children fail the standard false belief task

“Believe it, or Not? Explaining why children fail the standard false belief task” by Robert Thompson, Rice University.

ABSTRACT: It is not widely discussed, especially among philosophers of cognitive science, that children before the age of four can pass simpler versions of False Belief Tasks. There has been little discussion (and no consensus) about how to characterize the understanding these younger children manifest in these tasks. Success on these tasks, on the face of it, need not trouble the orthodox interpretation of the Standard False Belief Task (SFBT); these children simply understand the representational nature of belief, and hence, master the full-blown concept of belief, at an earlier age than commonly thought. Recent results have shown, however, that children as young as 13 months of age can pass such simplified tasks, and I think there are good reasons not to attribute mastery of the full- blown concept of belief to children at this age. Based on this evidence, I will argue that the abilities of these young children provide a serious challenge to the orthodox interpretation of the SFBT, and that we need a different analysis of the mindreading abilities of children at all of these ages. The major change that allows the child to pass the SFBT is not, I will claim, understanding the representational nature of belief. I propose an alternative analysis of the developmental data, stressing that understanding beliefs should be distinguished from mastering the full-blown concept of belief, and that the latter may involve capacities that go well beyond what has been described traditionally as aspects of ToM.

[Link to full text of article]
[Link to further info on PMS WIPS]

5 Responses to “PMS-WIPS 012 - Robert Thompson - Believe it, or Not? Explaining why children fail the standard false belief task”

  1. [...] Robert Thompson, Believe it, or Not? Explaining why children fail the standard false belief task of (from Brain Hammer PMS-WIPS) [...]

  2. Tad Z. says:

    Robert -

    Thanks for the interesting paper! I haven’t had the time to read it in the detail that it deserves, but some issues occur to me upon a first, light reading.

    *Which* false beliefs, specifically, do the infants in the study think the objects of their attributions token? Do they actually think - that adult believes the object is in the container? Do they have sufficient mastery of these concepts to attribute beliefs about objects instantiating them?

    Why call what the child attributes a *false belief*? I’m not sure what ‘belief’ means in this context. On one common, philosophical understanding of belief, any belief is compatible with any behavior, given sufficient adjustments to background propositional attitudes. So, it’s compatible with the behavior observed by the infants that the adults do not have the false beliefs that they attribute to them. Appropriate adjustments to background PAs can make the behavior entirely compatible with absence of these beliefs. In fact, in these experiments, this is actually the case: the adults do not have the beliefs attributed, but act as if they do, because they want (background PA) to test children’s capacity to attribute belief. My point isn’t that infants at this age should be able to understand this if they are to be attributed an understanding of belief. My point is rather that if they can’t even entertain the possibility that the same behavior can issue from different beliefs, then in what sense are they attributing *belief* rather than some other mental state?

    A final point. In characterizing the reasoning behind the experiment, you write that if “the child does not realize that her beliefs can be different from another’s beliefs, she should assume that since she saw the object being moved, the adult will know its current location as well.” This attributes to these children not only the concept of belief, but the capacity to attribute beliefs to themselves, and to compare these to those of others. Do the results of the experiment really support attributing such an elaborate understanding of self and other to such young children? It seems to me that there has to be a more deflationary interpretation. Surely such children aren’t able to wonder about their own beliefs, and whether they match those of others or not?!

    In general, for these sorts of reasons, I wish people didn’t refer to the perspective-taking skills revealed by such experiments as an understanding of (false) belief. There is surely some sensitivity to perspective - and I’m not sure how to characterize it. But to call it an understanding of (false) belief, seems to me a gross overinterpretation.

  3. Chase Wrenn says:

    JR-

    I’m glad somebody’s writing about these really early false belief tasks, and I’m glad you’re the one doing it.

    I think I have some similar (but not identical) misgivings to Tad’s, though. The data for the 13 monthers is far from giving unambiguous support to the view that they’re attributing false beliefs to adults. A more conservative interpretation would be that the children respond to the difference between cases in which people look where they put stuff and cases in which they do not look where they put stuff. It’s possible to explain what’s going on in terms of belief-attributions, but is it mandatory?

    Another thought also occurred to me towards the end of the paper. You suggest “that it is the improvement of Executive Functioning, Working Memory, and Linguistic Ability, along with a relatively stable understanding of belief, which finally allows children to pass SFBT.” Is there any thing you can say in support of this hypothesis? In particular, can you point to a detailed developmental model in which the roles of all these factors in passing SFBT are clear?

    -CW

  4. JR Thompson says:

    Hi Tad and Chase,

    Tad and I have now hashed some of this out in person, but here is something for those of you who weren’t down at the SSPP. I hope to add more elaborate responses later (after the central APA).

    My overall interest isn’t showing that young kids fully master, or deeply understand beliefs. As I note in the paper, my main beef is with those who think that what these kids really don’t get is that mental states (beliefs or some other more generic mental states) are representational. Whatever these young kids lack may be a full-blown concept of belief, but I don’t see how we should deny that they understand that mental agents (not just adults) as having a type of mental state which informs the agent about her environment and figures into goal-driven behavior in a way that closely parallels how beliefs work. I also admit that people may employ this more basic understanding of others in much of their mindreading (throughout life).

    So, my main point is to insist that what happens to kids at around 4yrs is not a major shift in understanding different perspectives or representations of reality. If these kids don’t understand what beliefs (or false beliefs) are, my opponents need to explain what else is missing. The orthodox picture just can’t be right.

    Now Tad makes a very interesting suggestion about what’s missing. This is quite clearly a thing which philosophers stress about belief and the other PAs, esp. after Quine. So, I don’t think that kids have to understand the multifarious relation between behavior and the conceptual scheme, holistically construed. First, this won’t help my main target–the defenders of the orthodoxy, since the earliest evidence I know of such an understanding emerges after kids pass the SFBT (this happens most clearly with kids understanding that another agent can express one emotion while feeling another). Second, I think this should count as something which is required to fully master beliefs, but is not required to understand their cognitive role in behavior. Hence, I think kids can get what beliefs are, and they would never get beliefs if they didn’t ignore this feature. I thank Tad for this suggestion–this is the sort of issue which I need to deal with in order to make sense of the understanding belief vs. grasping full-blown beliefs.

    Regarding Chase and Tad on the following sort of comment:
    In general, for these sorts of reasons, I wish people didn’t refer to the perspective-taking skills revealed by such experiments as an understanding of (false) belief. There is surely some sensitivity to perspective - and I’m not sure how to characterize it. But to call it an understanding of (false) belief, seems to me a gross overinterpretation.(from Tad)

    I think there is a parallel problem of describing these phenomena without a gross underinterpretation. The passage Tad cites is an example where there seems to be a mentalistic overinterpretation. But I don’t see why we need to be so squeemish about the mentalistic interpretation. This needs to be justified by showing how incomplete the deflationist accounts end up seeming.

    So, Tad doesn’t offer any deflationist reinterpretation of the data, but Chase does. Now, Chase suggests, about the 13month-olds
    A more conservative interpretation would be that the children respond to the difference between cases in which people look where they put stuff and cases in which they do not look where they put stuff. It’s possible to explain what’s going on in terms of belief-attributions, but is it mandatory?

    I suspect that Chase meant the 15month-olds. I don’t think that the deflationist accounts will work for the 13month-olds, and this is telling about how piecemeal the deflationist accounts turn out to be. This doesn’t mean that a deflationist account can’t be offered, and as Tad (at least) knows, I am quite uncomfortable attributing beliefs (in a robust sense) to kids this age. The problem for Chase’s suggestion is that the experiments for the 13month-olds is designed to undercut this sort of deflationist account (actually quite similar to the reply offered by Perner et alia in their response to the O&B results). What do the 13month-olds understand? It’s not just something about how people will search for things, based on where they left or placed those things. For the animated videos, the kid has to grasp where an agent will look for something, based on what it can currently see, what it has recently seen, and where it has recently gone. Now I don’t see how a non-mentalistic (and again, I’m fine if you deny this involves BELIEFS, in some grand sense, but please explain why these aren’t REALLY close to beliefs) scheme can explain all the expectations of these kids (even when they can barely talk!). Try to recast this as a behaviorist induction–how can the behaviors get explained? The kids would expect the creature to go to where the kid knows the favorite food is, or will expect the creature to go to the typical location of the food. What behavioral cues will allow one to deal with all the data? All the candidates seem more like gross underinterpretations than my gross overinterpretations. What we need is a better understanding of how a grasp of perspective figures into behavior. What I want to suggest is that the current view on this issue, the orthodox interpretation of the SFBT, just misses what’s really going on.

    As for Chase’s final point, I don’t have a good developmental story about how linguistic ability, EF, and working memory work. Part of the impetus for writing my paper was that I thought that too much focus was being placed on representational aspects of ToM, rather than these other areas. But, given my current leanings, I think that the revolution at age 4 is not so much an issue having to do with ToM. But, at bottom, I hope my paper inspires some theorists to take these sorts of alternatives seriously.

  5. Anibal says:

    I do not want to comment about what mechanisms are used by the toddler mind to acquire knowledge of the world, in this case the social world, wether his mind uses a representational system or a more intuitiive one based on visual description and emotion attunement, because it is a highly disputed issue, far to be resolved within the cognitive sciences in the next couple of days.

    But one thing explicitely stated in Thompson´s paper, is that the child is born with more cognitive capacities than previously recognized and this is apparently undisputed given the evidence (Dr. Anthony DeCasper).

    I would like to mention those disciplines called fetal cognition or neonatal psychology (Colwyn Trevarthen and DeCasper), which assumes some abilities, motivations and preferences in early states of development. In similar vein, it is the notion of “core knowledge” prooposed by Harvard psychologists E. Spelke and S. Carey, refering to unlearned knowledge about persons, numbers, places, objects and language when it definitely comes, that form the basis of the adult mind.

    I think Thompson´s paper follows this line of thought philosophically and it reperesent a new zeitgeist in mindreading debates, and what it says is that the ontology of toddler mind is not depleted like a “tabula rasa”, though empirical sources of knowledge shape it.