6 ms·

The curse of knowledge: Mistaking your beliefs for those of others

27 points by muon 17y ago

RiderOfGiraffes 17y ago
An interesting question, I wonder if anyone has actually done this and tested where Sally really did first look for her ball. With the rearranged receptacles, I'll bet some children really would look in unexpected places first.
- jerf 17y ago
  Well, when I consulted my theory of mind, it said that Sally would yell at Denise for moving things around and loudly demand "Where's my ball?! Where's my ball?!" without necessarily assuming that it's still in the box where she left it. Once the room has been moved around by a "hostile" agent, who knows where anything is? But that wasn't one of the options, apparently. I'm not trying to be funny; that really is my "best guess" as to the resulting behavior. The question itself brings in the hidden assumption that Sally will see her only choices as "look for the ball in one of four places" and in the real world the situation is much less constrained; running to Mommy is likely to result. I'm not complaining about the hidden assumption, either; all such hypothetical questions bring in hidden assumptions, and it's fair to use them here.
roundsquare 17y ago
People might be fooled by the fact that the bucket is where the box was before. I'm not sure how this plays in, but I wonder what they would do if people were told "the ball moves to the vase."
- calcnerd256 17y ago
  Exactly. Where's the control group?
apsurd 17y ago
I don't buy this. My first, seemingly "obvious" (to me) response was that Sally would look in the bucket. Not because its a bucket, but because its where the box was. I have no evidence for this, but isn't it natural of us humans to take spatial account of our surroundings. Don't people subconsciously remember where something is, like directions for example. You might not know the names of all the streets, but you just have a feeling, of where something is. Do we not ask: "Where did I put my keys!?" and not: "What did I put my keys, on, in, above, or below?!"
- jsankey 17y ago
  I'm not sure why you don't buy it on this basis. Your "obvious" response was evidently not that uncommon, as the first graph of adult responses shows (many more adults suggest bucket than basket or vase). The important part is the difference between this and the following graph, which shows that when adults know the ball is in the bucket (as opposed to any of the other containers) they are even more likely to answer bucket.
  - hughprime 17y ago
    I'm skeptical about this. The answer should be very obvious to any adult who has actually understood the question, so I'm thinking that any adult who doesn't answer "bucket" has just misunderstood the question (it's a little confusing to keep track of all these people and receptacles) and not that they lack a mental model of "other people's knowledge" the way that a three-year-old probably does.
    - gojomo 17y ago
      "The answer should be very obvious"? Huh? There's no obvious one right answer, though 'box' and 'bucket' are most defensible -- based on expectation of same container or same place. More adults answered 'box' and in reality I think what 'Sally' would do would be very dependent on the exact placement and the path she takes into the room, and her order of perceiving the containers. If she takes in the whole scene before deciding where to go/look, the box; if she moves to the expected place and then is surprised by the difference, the bucket. (This question in its first form may really be testing whether the answerer is primarily visual or kinesthetic in their modelling of other people.) But still, the essential point is the difference when people are told where the ball really is. 'Sally' still does not know, so a rigorously logical answer would be the same as the first case. (The additional information is irrelevant for predicting Sally's actions.) And yet, the additional information does cause a statistically significant difference in chosen answers. That means the answerer's mind, or model of mind, is leaky. Once 'primed' with info, they can't pretend they don't know it, or model the mind of someone who doesn't know it. Note in this case 'leaky' doesn't mean 'worse' (even though it is more like the younger child's answers). A straightforward policy of hewing to what is true regardless of what others know or how the knowledge was acquired is fine; it just isn't ideal for some social roles -- like lawyer or poker player -- that require a detailed understanding of others' limited information. So the real pro predictor, when asked "Where does Sally look first?", might ask in response: "How old is Sally? Does she more often pretend-play to be an athlete or a detective? What are her parents' careers?"
pbhj 17y ago
If Denise moves the ball to the bucket then perhaps putting balls in buckets is more common than putting balls in other containers. Sally being a child is likely to mirror Denises behaviours and therefore is probably more likely to place the ball in the bucket herself. She is thus more likely to look there than in other containers in which she herself would not place the ball. If one knows that the ball is in the bucket this reinforces the concept that balls go in buckets or reminds one that more often than not people put balls in buckets. Similar to the way that lots of people pick the same number in a given range or pick the same fruit when asked, etc., the most likely container for people to pick may be a bucket. People could be responding to an innate understanding of this fact; possibly without realising it. Speculation, but still a decent logical explanation IMO. A pre-experiment should be performed by asking people which vessel a ball should be placed in; the experiment in question could then choose only vessels with matching preference as ball receptacles as the options.

27 points by muon 17y ago

An interesting question, I wonder if anyone has actually done this and tested where Sally really did first look for her ball. With the rearranged receptacles, I'll bet some children really would look in unexpected places first.

jerf 17y ago
Well, when I consulted my theory of mind, it said that Sally would yell at Denise for moving things around and loudly demand "Where's my ball?! Where's my ball?!" without necessarily assuming that it's still in the box where she left it. Once the room has been moved around by a "hostile" agent, who knows where anything is? But that wasn't one of the options, apparently. I'm not trying to be funny; that really is my "best guess" as to the resulting behavior. The question itself brings in the hidden assumption that Sally will see her only choices as "look for the ball in one of four places" and in the real world the situation is much less constrained; running to Mommy is likely to result. I'm not complaining about the hidden assumption, either; all such hypothetical questions bring in hidden assumptions, and it's fair to use them here.

roundsquare 17y ago

People might be fooled by the fact that the bucket is where the box was before. I'm not sure how this plays in, but I wonder what they would do if people were told "the ball moves to the vase."

calcnerd256 17y ago
Exactly. Where's the control group?

apsurd 17y ago

I don't buy this. My first, seemingly "obvious" (to me) response was that Sally would look in the bucket. Not because its a bucket, but because its where the box was. I have no evidence for this, but isn't it natural of us humans to take spatial account of our surroundings. Don't people subconsciously remember where something is, like directions for example. You might not know the names of all the streets, but you just have a feeling, of where something is. Do we not ask: "Where did I put my keys!?" and not: "What did I put my keys, on, in, above, or below?!"

jsankey 17y ago
I'm not sure why you don't buy it on this basis. Your "obvious" response was evidently not that uncommon, as the first graph of adult responses shows (many more adults suggest bucket than basket or vase). The important part is the difference between this and the following graph, which shows that when adults know the ball is in the bucket (as opposed to any of the other containers) they are even more likely to answer bucket.
- hughprime 17y ago
  I'm skeptical about this. The answer should be very obvious to any adult who has actually understood the question, so I'm thinking that any adult who doesn't answer "bucket" has just misunderstood the question (it's a little confusing to keep track of all these people and receptacles) and not that they lack a mental model of "other people's knowledge" the way that a three-year-old probably does.
  - gojomo 17y ago
    "The answer should be very obvious"? Huh? There's no obvious one right answer, though 'box' and 'bucket' are most defensible -- based on expectation of same container or same place. More adults answered 'box' and in reality I think what 'Sally' would do would be very dependent on the exact placement and the path she takes into the room, and her order of perceiving the containers. If she takes in the whole scene before deciding where to go/look, the box; if she moves to the expected place and then is surprised by the difference, the bucket. (This question in its first form may really be testing whether the answerer is primarily visual or kinesthetic in their modelling of other people.) But still, the essential point is the difference when people are told where the ball really is. 'Sally' still does not know, so a rigorously logical answer would be the same as the first case. (The additional information is irrelevant for predicting Sally's actions.) And yet, the additional information does cause a statistically significant difference in chosen answers. That means the answerer's mind, or model of mind, is leaky. Once 'primed' with info, they can't pretend they don't know it, or model the mind of someone who doesn't know it. Note in this case 'leaky' doesn't mean 'worse' (even though it is more like the younger child's answers). A straightforward policy of hewing to what is true regardless of what others know or how the knowledge was acquired is fine; it just isn't ideal for some social roles -- like lawyer or poker player -- that require a detailed understanding of others' limited information. So the real pro predictor, when asked "Where does Sally look first?", might ask in response: "How old is Sally? Does she more often pretend-play to be an athlete or a detective? What are her parents' careers?"

pbhj 17y ago

If Denise moves the ball to the bucket then perhaps putting balls in buckets is more common than putting balls in other containers. Sally being a child is likely to mirror Denises behaviours and therefore is probably more likely to place the ball in the bucket herself. She is thus more likely to look there than in other containers in which she herself would not place the ball. If one knows that the ball is in the bucket this reinforces the concept that balls go in buckets or reminds one that more often than not people put balls in buckets. Similar to the way that lots of people pick the same number in a given range or pick the same fruit when asked, etc., the most likely container for people to pick may be a bucket. People could be responding to an innate understanding of this fact; possibly without realising it. Speculation, but still a decent logical explanation IMO. A pre-experiment should be performed by asking people which vessel a ball should be placed in; the experiment in question could then choose only vessels with matching preference as ball receptacles as the options.