Preschool children rarely seek data when observation and testimony conflict | Hacker News Reader

Preschool children rarely seek data when observation and testimony conflict

srcd.onlinelibrary.wiley.com

154 pointsDocFeind83 comments

> Elementary school children significantly increased their exploration of the dolls when their intuitions had been contradicted as compared to when they had been confirmed, frequently picking up the smallest and the biggest doll concurrently to compare their relative weight—a direct test of the claim they had been given. Preschool children rarely engaged in this behavior, whether their intuitions had been confirmed or contradicted.

Interesting study. Good to make clear that this does not apply to older children. Older children will verify any surprising claims. On the other hand they suffer, like adults, of confirmation bias. If the information fits their expectation they will not check is veracity. I guess that confirmation bias is just an efficient way of learning about the world, do not waste time if things seem to fit your current understanding of the world.

To see a creature with not enough confirmation bias watch a horse if it encounters something even slightly usual. It will shy away and or investigate or refuse to move until you have investigated the slightly oddly positioned leaf on the path.

If we did indeed check everything we nevertheless expected to be as usual we’d never get anything done and would get stuck in mega loops.

To hilarious effect with regard to llama competitions at the state fair. Very serious twelve year olds (god bless ‘em) leading their llamas through a rigorous course of “touch your nose to a mirror of death”, “walk the gauntlet of pool noodles”, “step across the tinfoil floor of doom”, “remain calm under the umbrella’s predatory gaze”.

Horses spooked by unexpected street markings https://www.youtube.com/watch?v=TMyf6ewi7E8

...and this is why road markings have legal standards in many countries.

Interesting. Cows won't step over painted parallel lines if they have encountered a cattle guard before.

Where a cattle guard is parallel pipes in the road at a gate opening meant to keep them in...their hooves can't negotiate the spaces between the pipes.

Real one: https://www.conteches.com/Portals/0/Images/US%20Forest%20Gri...

Painted one: https://i.stack.imgur.com/gT5vU.jpg

The painted one is sorta funny, such a low tech solution. I recently came across this interesting tidbit: "Cattle grids are generally useless for containing goats. However, a Texas Highway Department official reported that adding three 20-inch (51 cm) painted stripes—arranged yellow, white, yellow—on the road in front of a cattle grid deterred goats from approaching or crossing the guard."

https://en.wikipedia.org/wiki/Cattle_grid

My brother did civil war reenacting pretty seriously (he had a bit part as a union soldier running across the battlefield and getting shot dead in the film Gettysburg, among other things) and he told me that a horse would get spooked by a coffee mug if it hadn't seen one before.

It gives you a whole new appreciation for HEMA and what it took to train destriers.

I guess the utility of a novel thing to a herd, prey, plains animal is mostly likely about zero. The danger of the thing is higher.

Horses also need to be desensitized to new objects twice: once on their right eye and once for their left.

That is an interesting observation. As prey rather than predator, however, a horse’s response has, more or less, been validated by the survival of the genus.

I would guess that a wild horse or mustang would not be frozen in place by such things, suggesting, perhaps, that domestication has created a conflict between instincts, which is sub-optimal for its rider, but not necessarily for the horse.

With regard to the general point about checking everything, you are correct, and it takes a certain amount of judgement to doubt one’s intuitions without being paralyzed by doing so.

Mustangs behave pretty much the same as other horses when confronted with something scary. They're not really any different to work with than any other unhandled horse. Mustangs are not wild horses, btw. They're feral horses descended largely from the horses of the Conquistadors.

So much for my guess! BTW I am well aware that mustangs are not wild horses, which is precisely why I made the distinction in my original comment. When you say "any other unhandled horses", are you including truly wild horses?

There arguably aren't truly wild horses anymore. The closest, Przewalski's horse, only exist in the wild as a population reintroduced from captivity. The captive population that they were reintroduced from very might well have been a more docile subset of the subspecies, making your question difficult to answer.

> I guess that confirmation bias is just an efficient way of learning about the world, do not waste time if things seem to fit your current understanding of the world.

Interesting observation - in this light, confirmation bias could almost be considered a variant of Occam's razor.

yep, that's the simplest explanation.

Check lists of other acknowledged biases, such as putting too much weight on earlier presented information

Occam's razor is a variant of confirmation bias.

Could it be that children lacked the knowledge of how to determine whether or not the figurines worked, or believed they were not allowed to touch them? Neither interpretation seems plausible. Recall that children had just seen the naïve experimenter place the figurines on the box, and were asked to sort the figurines by this naïve experimenter, not by the apparently knowledgeable experimenter who told them about which figurine worked. In principle, the fact that the person who asked children to sort the figurines had tried them out on the music box as children watched, and also implied that exploring the figurines would provide information about how the music box worked by saying “I don’t know how it works, but it looks like you can put these pieces on this thing here [placing a hand on the center of the music box]. I wonder what will happen if I do that!”, should have primed children’s exploration if they conceptualized such exploration as informative. Furthermore, the naïve informant left the table and sat faced away from the child during the task, attempting to remove potential social pressure not to question the informant’s testing. However, given that testing the figurines would have been audible to the nearby naïve experimenter, future studies may be useful to assess whether additional children would have tested the figurines if left in total privacy.

I think that children might well have felt they weren't supposed to test the figurines on the music box itself, and this explanation doesn't completely satisfy me. The setup makes it seem like some kind of game, and testing directly on the music box might intuitively seem like cheating.

Reading that description, and knowing preschoolers pretty well, I'd suspect that the experimental setup is far too complicated for making any claims about children would interpretation the experimenters (fake!) behaviors

I think they should have done this with an incentive. Basically, don't ask "If I want to make the music box play, which figurine should I use?" But tell the children before the sorting task that if they make the music box play you'll give them a marshmallow. Without an incentive the kids may just not care enough to investigate.

Exactly this. The concluding claim from the paper;

> Thus, this study is consistent with the proposal that the majority of preschool children do not seize opportunities to engage in structured empirical investigations with the explicit purpose of resolving conflicts that arise from surprising testimony

completely ignores the bioeconomics of carrying out “structural empirical investigations”. It is more expensive to carrying out investigations than to take a best guess. In the presence of tons of other cognitive tasks they could be doing, they have no obvious reason to carry out this particular and ostensibly boring investigation. Knowing the natural curiosity of kids, they have a high baseline of investigative behavior.

I was thinking this at first too, because if there's no reinforcement or punishment, then the instructions to sort the figurines based on whether or not they will activate the music box is ambiguous — should you sort them such that other people will think they're sorted correctly or so that they actually are?

But now I'm less sure. The adult giving the instructions to sort the figurines is the one who was presented as not knowing how the music box worked. They were the only adult in the room and remained in the room while the child performed the task, which would suggest to the child that they would be the one evaluating them. Since the person evaluating the child's performance was not the one giving the misleading testimony the two possible interpretations of the goal of the task shouldn't give conflicting behavioral prescriptions.

Yeah, I think the design of the experiment is good to catch this potential weakness. If it was the knowledgeable experimenter instructing the children, that could really confuse things. I think it's still potentially a source of confusion, but less of one.

If you imagine that instead of sorting 4 figurines they had to sort 400, then it would be easy to see that the kids aren't going to test all of them. Too boring. They may not even go to the trouble of testing a sample of the figurines. The would need some incentive to drive them to do a bit more work than the bare minimum.

If I ask my son to clean his room, and I return to discover he's done a poor job, I will not think that he doesn't understand how to clean his room. Similarly, if these kids are asked to sort figurines and they just go by what they're told (or what they observed) without testing, that doesn't necessarily mean they don't understand that testing might be useful.

I think making the music box play is supposed to be rewarding in itself.

But 60-70% of the kids didn't even use the music box when they could have. Not a great reward.

Better yet, threaten to punish them. It's been shown that negative reinforcement is more effective than positive

Common mistake: Negative reinforcement is when you take a negative stimulus away to reinforce the desired behavior.

What you're describing is positive punishment.

Also nicely explained in TBBT: https://youtu.be/gLp_aLMm5qQ

well it shouldn't. makes much more sense to use negative reinforcement as a term to describe a reinforcing stimulus of a negative nature.

optimal terminology has never been our strong point as a scientific community.

makes more sense at first glance maybe, but when you dig into it more it's perfectly logical.

"positive" refers to adding stimulus, "negative" refers to removing stimulus.

"Reinforcement" is anything that increases the occurrence of the behavior, "punishment" is anything that decreases the occurrence of the behavior.

Perhaps we should have used “stimulating” and “depriving”. This reminds me of the whole Type One and Type Two error thing.

"additive" and "subtractive"

That doesn’t make sense at all. That interpretation would make “positive reinforcement” and “negative punishment” redundant, and it would make “positive punishment” and “negative reinforcement” contradictory. It’s very difficult to effectively reason about modes of operant conditioning when you throw that much nuance out of the window.

I don't see why. a) the word "punishment" is misleading, since not all negative stimuli are punishments in nature. b) if removing a stimulus is reinforcing behaviour in some way, then it is reinforcement.

The argument isnt to reverse the meanings of existing labels containing the word punishment. The argument is that the four categories should be, e.g.: 1. "Positive": Reinforcement of behaviour from positive stimulus 2. "Antipositive" (or similar term): Reinforcement from reversal of positive stimulus 3. "Negative": Reinforcement of behaviour from a negative stimulus 4. "Antinegative": Reinforcement from reversal of a negative stimulus.

The thing is punishment is not reinforcing behavior, it is prohibiting behavior.

Reinforcing means increasing the likelihood (by adding something liked or by removing something disliked) while punishment refers to reducing the likelihood (by adding something disliked or removing something liked).

If you look at it from a logical point of view both reinforcing stimuli are positive (in the sense of good for the subject, i.e. directly good or not bad) while punishing stimuli are both negative for the subject (in the sense of bad for the subject, i.e. directly bad or not good).

The language problem and common mistake comes from the usage of positive and negative with a different meaning from the usual meaning in the context of behavior. Thinking of it like additive and subtractive is more helpful. But making the meaning of reinforcement less specific isn't really helpful.

I think we'll agree to disagree, since this is literally playing on words.

To me "reinforcement" alludes to a reinforcement of behaviour, not likelihood of action vs inaction. So it's a case for increasing the specificity of the word, not decreasing it.

If every time you try to open the door you get an electric shock, this is a "negative/punishing" stimulus that reinforces a particular behaviour of avoiding doors. Similarly if you used to get a treat every time you opened the door which then reinforced a door-opening behaviour, and then you suddenly stop giving treats, then this is the removal of a "positive/rewarding" stimulus, which 'weakens' the particular, previously reinforced behaviour of opening doors.

At the very least we should be talking about 'reinforcement vs weakening', 'positive/rewarding stimuli vs negative/punishing stimuli', and 'application vs removal of stimulus'.

So, linguistically speaking, using the term "negative reinforcement" to mean "weakening by removal of a positive stimulus" rather than "reinforcement via application of a negative stimulus" is simply misleading, and bad choice of terminology.

I do agree that the terminology isn't ideal and easily misleading, that's why it's so commonly used incorrectly. And weakening might be more fitting than punishment.

> So, linguistically speaking, using the term "negative reinforcement" to mean "weakening by removal of a positive stimulus" rather than "reinforcement via application of a negative stimulus" is simply misleading, and bad choice of terminology.

Using your terminology "Negative reinforcement" actually means "reinforcement by removal of a negative stimulus". "Weakening by removal of a positive stimulus" is the description of negative punishment. The crux of the problem is that positive and negative are not a descriptive property of the stimulus in this context, but just synonyms for additive/subtractive and I agree that this is misleading.

It's wrong to threaten to punish a child just for the fun of it, even if "fun" is cloaked in a scientific study.

Nobody said anything about fun. Punishment can be done easily and ethically in child studies. It can be as simple as starting them with a marshmallow and telling them you will take it away if they are wrong.

1. Sweets and food should not be used as reward and punishnent in the first place. So that not eating them is not emotionally punishment, but just healthy thing to do.

2. Cue meltdown in many kids with that tactic - you bring up ecpectation and then crash it.

3. Have you people even interacted with kid? What it is with adults who seek to design punishment into situation where there is already compliance or which is about unimportant bs?

> 1. Sweets and food should not be used as reward and punishnent in the first place. So that not eating them is not emotionally punishment, but just healthy thing to do.

That’s a nice thought, but millions of years of co-evolution with plants using animals for seed dispersal trumps a few thousand years of post-agricultural revolution excesses. Maybe in another million years we’ll lose our genes favouring sweet foods, who knows, but until then candy is a safe bet.

> 2. Cue meltdown in many kids with that tactic - you bring up ecpectation and then crash it.

Unlikely given that the experimenter will be a total stranger. There will be no established incentive structure to use tantrums for psychological manipulation as with the child’s parents.

> 3. Have you people even interacted with kid? What it is with adults who seek to design punishment into situation where there is already compliance or which is about unimportant bs?

That can be said about any area of study. Even a study of a supposedly solved problem or “unimportant bs” can lead to great insight.

Im not sure if we are taking about the same topic, but in case you are

>1. Sweets and food should not be used as reward and punishnent in the first place. So that not eating them is not emotionally punishment, but just healthy thing to do.

Kids like treats, have you met one?

>2. Cue meltdown in many kids with that tactic - you bring up ecpectation and then crash it.

Preschoolers can understand conditional rewards and penalties and play simple games. We arent talking about toddlers here.

>3. Have you people even interacted with kid? What it is with adults who seek to design punishment into situation where there is already compliance or which is about unimportant bs?

Not sure what you on about here. What situation and what compliance are your referring to? We are taking about the design of a human behavior study, not how people should raise their kids.

> Kids like treats, have you met one?

My comment was about what you attitude you teach them about sweets. Nothing in it implied they dont like it. They still should not be used as reward.

Not having sweets is not punishment. It is just healthy behavior. You dont eat sweets as reward for being good either. Being good dont imply ability to eat more sweets.

> Preschoolers can understand conditional rewards and penalties and play simple games

Preschoolers have meltdowns over those super regularly. They actually completely sux at games with rules over rewards and punishments, because they sux at hangling the disappointment.

Have you worked with preschoolers, ever?

> Not sure what you on about here. What situation and what compliance are your referring to? We are taking about the design of a human behavior study, not how people should raise their kids.

The kids did as scientists asked them. That is compliance - there was no issue of kids ignoring doll or going to play chase.

> My comment was about what you attitude you teach them about sweets. Nothing in it implied they dont like it. They still should not be used as reward.

Sure, but it’s hardly the time and place to grandstand about healthy eating, is it? The point is to find an ethical and less confrontational alternative to something like hitting the child with a ruler. You’re welcome to suggest a better idea.

> You dont eat sweets as reward for being good either.

Sure, in an adult study it would probably make more sense to have a monetary reward, but with children it’s better to use something less abstract.

> The kids did as scientists asked them. That is compliance - there was no issue of kids ignoring doll or going to play chase.

The idea is to figure out whether the poor performance was because they weren’t motivated to make an effort or because they were just dumb. It’s only natural for humans to put in only the minimum effort on a pointless task with no reward or risk of punishment. I certainly can’t think instance in my child- or adulthood where I made effort for no reason.

>The kids did as scientists asked them. That is compliance - there was no issue of kids ignoring doll or going to play chase

This will be my last response because you clearly don't understand the OP study or the discussion in this thread. The study looked at a problem solving task with conflicting instructions, and measured the ability of children to complete the task correctly.

acituan and ALittleLight raised valid points that rewards and punishment could change the study results. For example, maybe children would be complete the task better better if they were punished or rewarded if they complete the study wrong for right. My point is that while it would be unethical to beat or harm a child for completing the task wrong, there are ethical ways to include some other forms of punishment.

It seems like you have some pet social and behavioral issues, but I find your poor grammar and communication skills trying, and have no interest in discussing them further.

Source?

Just preschool children? Adults on HN rarely seek data when observation and testimony conflict.

The example I have observed over and over on HN is from people on the US/Canadian west coast and property prices versus availability. They shrug off any need for data by talking about high school economics. When you point out that this isn’t data others will chime in confusing apartment leases for property ownership.

I've seen grown adults rarely seek data when observation and testimony conflict

Exactly. The whole idea of structured experimentation (basically science) is relatively new and most people don't spontaneously do it unless trained to do so. Just look at the wide variety of easily falsifiable beliefs that still find widespread acceptance.

It’s a good heuristic. You don’t have enough data to make good generalizations, and can quickly acquire knowledge from people who (evolutionarily, else the heuristic would not be instinctual) in general won’t lie to you.

And then at around six you push the reset on experiential learnings, but now with a pretty good base dataset.

Unless I am missing it, they don't report the results by gender, but they do by age. Is this suggesting that the results were identical across gender?

> Is this suggesting that the results were identical across gender?

I suppose you mean sex, since using genders for preschoolers would be ridiculous.

Secondly no, it's not suggesting that. It's simply suggesting they may not have grouped their results by sex, maybe because that would have made the groups too small to yield useful data, or because they simply weren't interested in that.

As an aside, you shouldn't post-factum partition your data by characteristics you did not set out to compare, because then you might try height, intelligence, hair color, etc. until you find some characteristic that yields differences of "statistical significance" even though you've just been trying them all until you found something that differentiated some groups by random chance.

I mean, if a part of the experiment wasn't "inspect the childrens' genitals" I think it would me more accurate to call the split a gender split rather than a sex split even if it results in the same groups.

Why would age (3 groups), which they did partition on, be statistically significant but not sex (2 groups, assuming no intersex)?

That's just not what I said.

>maybe because that would have made the groups too small to yield useful data

How would 2 groups be "too small to yield useful data" when 3 groups yield useful data.

Because you would be subdividing again, which yields 6 groups.

Also read my last paragraph about 'playing' with your data like that.

>Because you would be subdividing again, which yields 6 groups

No, it would be 2 groups if the single partition was sex. It would only be 6 groups if they also wanted to slice each group again by age, which is not what I'm asking. The 2 groups are not "too small" if 3 groups are not "too small."

>Also read my last paragraph about 'playing' with your data like that.

I did, and I think that's valid. However, it has no bearing on my point about the size of the groups.

Maybe they were all boys.

The sex split is within the first sentence of the study Abstract. It was 54% girls.

I don’t think adults are much better. “Listen to the experts” and all that.

You are more likely to obtain the correct result from a credentialed expert than doing your own research. This should be clear in the ever growing landscape of unthinking people using “research” as a proxy for confirmation bias.

> You are more likely to obtain the correct result from a credentialed expert than doing your own research.

That clearly depends both on your and the expert’s judgement, and I’d say the variance of both is high.

Give me 1 day to cram about any disease that's remotely obscure, and I bet you I can out-test any general practitioner.

Yeah, for most things the hard part is building the base knowledge so you know what information is trustworthy. After that, intensity beats generality on almost anything that has documented knowledge.

I agree with you.

Surely unless you are a doctor yourself, you cannot perform the clinical trials needed to arrive at an objective conclusion in just a day's time. So you must be relying on studies published by--gasp--experts!

Phrased differently: Give a general practitioner a day's notice to read up on an obscure disease, and they'll surely run circles around YOU.

That’s the problem, actually. For a personal problem, you will easily dedicate that day. The GP isn’t going to, though.

So your choices are:

- Dedicated research by you

- Off the cuff answer by GP

The universe where you get dedicated research from the GP doesn’t exist.

This means that nearly any intelligent person walking into a consultation with a GP for a condition they have likely knows more than the GP and only needs the credentialed expert for prescription and satisfying insurers.

That depends on how skilled you are at performing research, and whether or not you have enough "base truth" knowledge in the domain to filter out low quality information.

9 times out of 10, though, I think your observation is correct. The people doing "research" get their information from YouTube documentaries linked to Facebook groups, and have very little practical understanding of epidemiology/nutritional content of meat/how child sex rings operate.

So, pre-school children are more likely to investigate a conflict if the testimony is provided before the observation than if it is provided after. Basically, if you want to avoid confirmation bias you'll need to form your own opinion ahead of the event.

No, neither this nor opposite was observed if i read correctly

not just children

Could this have more to do with the fact that pre-school aged children have the attention span of a puppy because their brains aren't fully developed yet?

Or perhaps because the intuitions of pre-school-aged children are still forming; there's a lot they don't know, and the world is new and exciting, so scepticism would mostly just rob them of learning opportunities.

From what I remember of being that age, I think I just didn't notice things. As an adult (and as an older child) I often get a "that doesn't seem right" feeling/intuition. As a very young child I didn't know what normal was or what the important parts of experiences were so I was much more open to things and just went with them.

I've got used to relying on my “that doesn't seem right” feeling for determining whether I know stuff. In fields where I know enough to not feel ignorant, I say all sorts of things that don't feel wrong, but they actually are. (When talking about linguistics, for instance, I am confidently incoherent. I know this, and yet the next time I want to say something about linguistics, I do not know that I am speaking garbage.)

I don't think pre-school aged children even know where to look for the data yet.

I think they do: Ask their parents or teachers.

But remembering it long enough to get away from the person who told them the fact and then ask a different person... That's another story. I'd think far too much would be happening for them to bother remembering to ask.

Being able to look it up independently of another human makes it a lot easier to verify information immediately, so I'm not surprised that people who can read are more likely to verify info than those who can't.

In this test the kids are sorting figures that do or don't activate a music box. The box is left next to them so they could use the box to test the figures prior to sorting them or they could rely on what an experimenter told them (white figures activate the box) or what they observed when a different experimenter was actually using the toy. What the title means by "seek data" is "Did the kids try the figurines on the box themselves?"

They also get a ton of data about the world through free play, especially outside in nature, but they experience it as play, not data gathering.

Yet, this describes plenty of older people as well (teens/20s).

I know this is a popular preconception and it feeds into all our biases about lazy youth that just needs to get off our lawn, but any teenager will readily out-think a preschooler.

whoooaa, you're being very generous there. "any teenager" is doing a lot of heavy lifting. i have the unique experience of being one at one point. of course i'm being sarcastic, but some of the dumbest shit i've ever done (hopefully will ever have done) was as a teenager. a toddler does things because "they don't know any better". a teenager does things even though "they should know better". i've seen medical people talking about the brain still not fully finished baking until the 20s. so with that unfinished gushy frontal lobes and all of those raging hormones with that natural teenage angst, it's a wonder we survived as a species at all ;-)

There's a difference between wisdom and intelligence. Teenagers tend to lack the former, not the latter.

Reply on news.ycombinator.com