So let's talk a little bit about intellectual humility and belief revision.
We've established that explanation has a unique benefit on learning.
An explanation is part of this process,
and we know that revising beliefs or revising
explanation is at the core of intellectual humility.
After all, in order to be humble about what you know,
you have to be willing to be wrong.
You have to be willing to revise your beliefs.
Now, examining constraints on explanation I think
provides insight into the process of belief revision.
What kinds of context and information motivate children to revise their beliefs?
And this is a crucial part of understanding how
the development of intellectual humility proceeds.
So we know from Karl Popper
that observation is always selection.
It means a chosen object, a definite task,
an interest, a point of view, a problem.
When we generate an explanation,
we're always looking to explain something in particular.
We're not interested in explaining everything.
So a big part of what I want to understand and what I've
studied is what motivates children to explain,
what piques their curiosity?
This is crucial to understand how to get children to revise their beliefs.
We have to get them to engage in understanding in the first place.
We know that explanation is a selective process.
We also know that like adults,
children aren't interested in explaining everything.
They're not curious about everything.
Children are much more curious about some things than others.
And if we want to optimize learning,
it follows that we should know a lot about what children are curious about,
what piques their curiosity and their interest.
And examining what motivates explanation provides insight
into the process of scientific reasoning and belief revision.
So let's talk a little bit about what kinds of events motivate scientific reasoning,
and in particular motivate children to construct explanations.
We know that children explain events that have the potential to teach them something new.
Some particular inconsistency with
prior knowledge triggers childrens' explanatory reasoning.
Children are much more interested in explaining an
inconsistent versus consistent outcome.
And the quality of their explanations, sophistication of that explanation
for an inconsistent outcome is greater than for a consistent outcome.
We've also shown that children's explanations guide their exploratory behavior.
So the way children explain outcomes influences the sorts of behaviors that
they engage in to test whether their explanations are accurate or not.
We know that children preferentially
explore inconsistent outcomes that they have explained.
So there's a dynamic process between explanation and further exploration.
So I think explaining inconsistency is in fact a really crucial mechanism for learning.
Inconsistent outcomes pique children's curiosity,
motivate children to generate explanations which subsequently
guide their exploration and hypothesis testing behavior.
This generates new information which feeds
back into the kinds of explanations that they generate,
and the process by which they revise
their explanations in response to inconsistent or new information.
So to do this experimentally,
we control children's expectations and beliefs.
Let me give you an example of how we've done this using novel stimuli- these light boxes.
So to do this what we've done is we present children with outcomes that are
consistent with prior knowledge as well
as outcomes that are inconsistent with their prior knowledge.
We present those outcomes simultaneously to
children and then we ask them why. Why did that happen?
To do this, we use a novel task with novel causal properties and categories.
Now the benefit of this design is that we have
a number of different measures of children's explanation.
We can measure what they explain first,
the content of their explanations,
and how their explanations influence
the kinds of exploratory behaviors that they engage in.
So here's an example of an experimental task that we've used.
We use these novel light boxes,
and novel objects with novel labels.
This is to introduce new information to
children and to prime what their expectations are.
So first we have a training trial.
The TOMA training trial,
and here we put a 'toma' on each of the boxes individually.
And what you see is that tomas appear to activate the box- they turn the light box on.
Next, we have a training trial for 'blickets'.
And here we placed the blicket on each of the objects, again, independently.
And as you can see the blickets do not appear to turn the light box on.
Next, we have a trial where we simultaneously place one toma and one blicket on each box.
As you can see, the toma activates the box exactly as you would expect.
And the blicket does not activate the box.
And we asked the child, why did that happen?
Then there's another trial where once again a toma and a blicket
are placed on each of those boxes simultaneously.
And what you see here is that the blicket does not activate the box,
again totally consistent with what you would expect,
but in this case the toma does not activate the box,
which is inconsistent with your expectation.
And once again we ask the child,
why did that happen?
And following that explanation,
we give them an opportunity to engage directly with the objects.
So once again, we can analyze what they explained first,
the content of their explanations,
and the kinds of hypothesis testing behaviors that
they engage in and their attempt to understand why
that inconsistent outcome doesn't appear to work.
Nothing. This one lighted up,
and this one did not. So that means... What's making this light up?
And what we found is that explanation guides exploratory behavior.
Children preferentially explore inconsistent outcomes they have explained.
Children also are more likely to first explain that inconsistent outcome suggesting
that inconsistency piques children's curiosity and motivates their explanations.
Always, because this needs to be like this.
And this needs to be like that. That's why. Why are the boxes on the wrong way?