Keyboard Shortcuts?f

×
  • Next step
  • Previous step
  • Skip this slide
  • Previous slide
  • mShow slide thumbnails
  • nShow notes
  • hShow handout latex source
  • NShow talk notes latex source

Click here and press the right key for the next slide.

(This may not work on mobile or ipad. You can try using chrome or firefox, but even that may fail. Sorry.)

also ...

Press the left key to go backwards (or swipe right)

Press n to toggle whether notes are shown (or add '?notes' to the url before the #)

Press m or double tap to slide thumbnails (menu)

Press ? at any time to show the keyboard shortcuts

 

Dual-Process Theory of Cognitive Development

Metacognitive Feelings and Development

Knowledge of Objects & Minds

When and how do humans first acquire knowledge of simple facts about particular physical objects and their causal interactions, or about what another person believes?
In this presentation I will introduce a very simple dual-process theory of development; identify a problem for this theory; and argue that solving this problem depends on recognising a role for metacognitive feelings in development.
My aim is to support the view that metacognitive feelings may be important in the development of knowledge of objects and minds.
Infants were just under 3 months in this experiment.
I mention this paradigm because it ancient history (1992) and has since been much replicated.

Spelke, Breinlinger, Macomber, & Jacobson (1992, p. figure 6 (part))

Spelke et al. (1992, p. figure 7 (part))

Hood, Cole-Davies, & Dias (2003)

There is a surprising discrepancy between infants’ performance on this task and their performance on object search tasks ...
In this scenario, a cylinder with a smiley face rolls down a ramp and stops behind one of the doors in the screen. You can tell which door it stops behind because there is a barrier sticking up above the screen.
Your task is to open a door and retrieve the cylinder, and you may open one door only. 2.5-year-olds appear to open doors at random (often having a favourite door), and slightly older children opened a door adjacent to the barrier but had no preference for the door on the correct side of the barrier (Berthier, De Blois, Poirier, Novak, & Clifton, 2000; Hood et al., 2003).
But what happens if we recast the experiment as a violation-of-expectations task? In this case, subjects merely observe as the experimenter removes the cylinder from one or another door. Of interest is whether they look longer when experimenter removes the cylinder from behind the wrong door (Hood et al., 2003) or when the experimenter opens the correct door and there is no cylinder (Mash, Novak, Berthier, & Keen, 2006). 2-year-olds look longer in both conditions (Hood et al., 2003; Mash et al., 2006).

Hood et al. (2003)

So we see the same initially puzzling pattern in the case of knowledge of objects and causation.
If you look just at this pattern, you might suspect that observing R1 gives us a better, more sensitive and less noisy measure of infants’ competence.
And that would be reasonble except that ...
You can also get the reverse pattern using other tasks!
Importantly, we appear to find much the same initially puzzling pattern in other domains of knowledge. For example, consider mindreading ...
Here is Jason who wants to retrieve a red robot from under of these boxes.
When these windows light up, he will reach through one of these windows to retrieve it. In fact the red robot is in the right box. But Jason has a false belief about it’s location: he falsely believes that it is in the left box.
We can measure whether you take Jason’s belief into account in two ways. We can observe where you look in anticipation of Jason’s action when the windows light up (initial proactive gaze). And then, a few seconds later, we can ask you where you think Jason will look first.
Extending groundbreaking work by Clements & Perner (1994) and many following them, Low and colleagues have done several variations on this, with 3- and 4-year-olds as well as adults and found just the pattern here: an interaction with age.
At this point some are tempted to claim that this pattern of findings is not anomalous at all. It’s simply a consequence of the fact that the some responses provide a more sensitive measure of competence, or that other responses require some feature extraneous to the task.
But this claim is less persuasive given a further piece of the evidence. For some carefully matched tasks, responses like initial proactive gaze and looking time indicate very poor, below chance performance. This is hard---not impossible, but hard---to reconcile with the claim that the these responses simply provide more sensitive measures. It is also hard to reconcile with the claim that the interaction with age is due to extraneous features introduced when responses like verbal prediction are involved.

Low et al, 2014 figure 2

(Incidentally, I admint that I’m using a made up figure to illustrate a theoretical point. But it’s actually not too far from how real data from even a single study sometimes looks.)

Q1

Why do different responses reveal different developmental patterns?
 

So this is my first question: Why do different responses reveal different developmental patterns?
To date people have mostly regarded the interactions with age as a quirk of methods or of data.
They have therefore attempted explanations that are specific to one domain. They treat the interaction in the case of object cognition as one thing, and the interaction in the case of mindreading as a separate thing.
But the interactions are so pervasive that this cannot be satisfactory.
I propose to treat the interactions as informative rather than mere quirks. This gives us an interesting answer to our question.

dual-process theory of cognitive development

There are two (or more) processes.

One process is faster than another.

Faster processes are relatively unchanging,
slower processes change across development through learning.

Faster processes dominate anticipatory looking,
slower processes dominate verbal responses.

So if a dual-process theory is right, the development of mindreading looks something like this.
The early-developing,\footnote{The term ‘early-developing’ is potentially misleading. According to the dual-process theory of development, all processes may be operational from early in development. The difference is just that some do not undergo significant developmental change. I should really have used a term like ‘developmentally unchanging’.} relatively automatic process is constant over development. Whereas the less automatic process starts off with an extremely crude model of minds and actions and only gradually accrues sophistication.

Q1

Why do different responses reveal different developmental patterns?
 

How does the dual process theory help us with understanding the initially puzzling pattern of development?
I asked, Why do different responses reveal different developmental patterns?
It may be because the process dominating R1 changes relatively little over development, while the process dominating R2 undergoes gradual change over months or years.
In essence, the idea is this. If the processes dominating responses R1 and R2 are different, then the fact that these responses reveal different relations between age and performance simply indicates that the two processes develop differently.

?

predictions

I think this is a nice story. But of course there are other nice stories you could tell. What are the predictions of the dual process theory of development?

Predictions:

[1] Where infants can track objects or beliefs, in adults there will be two distinct processes ...

[2] ... and one adult process will have features in common with the infant process.

The core of a dual-process theory of development generates two key predictions. about relations between adults’ and infants’ performance:% \footnote{See further Grosse Wiesmann, Friederici, Singer, & Steinbeis (2016).} \begin{enumerate} \item Where infants can track beliefs, there will be two (or more) distinct processes in adults ... \item ... and one adult process will share features with the infant process. \end{enumerate}

Ex: tracking briefly occluded objects

Let me illustrate this by referring to some hopefully quite familiar work on infants’ abilities to track briefly occluded objects.

Occurs in 4-month-olds ...

... and we can distinguish two object-tracking processes in adults (e.g. Mitroff et al, 2005).

In adults, one object-tracking process is subject to a signature limit...

... which is also a limit of infants’ object-tracking (e.g. Xu & Carey, 1996)

To generate further readily testable predictions, we need to go beyond the schematic part of a dual process theory of development, the part which is common to dual process theories of memory and of physical cognition. To generate further predictions we need to fill in the the domain-specific parts of the dual process theory.
I won’t cover it in this talk, but much work along these lines has already been done in the domain of minds, actions and physical objects.
[table: domain / mechanism-or-model / signature limit / ref. for evidence]
domainmechanism/modelsignature limitsource (e.g.)
objectsobject indexesfeatural cuesXu & Carey, 1996
mindsminimal theory of mindidentityEdwards & Low, 2017
actionsmotor processesown abilityKanakogi & Itakura, 2011

Q1

Why do different responses reveal different developmental patterns?
 

So far I’ve suggested that (a) a dual process theory of development can help to explain an initially puzzling interaction between age and response type concerning performance, and that (b) by linking infants’ and adults’ performance, a dual process theory of development generates testable predictions.
[repeat:] In essence, the idea is this. If the processes dominating Response 1 and Response 2 are different, the different relations between age and performance these responses reveal simply indicate that the two processes develop differently. The process or processes dominating Response 1 changes relatively little over development, while the process dominating Response 2 undergoes gradual change over months or years.
\textbf{It’s not my aim in this talk to defend a dual process theory of the development of object cognition or mindreading. (I think we should just wait and see how its predictions hold up.). Instead I want to suggest that any such theory faces a significant challenge ...}
The challenge arises from the various ways in which infants manifest their abilities to track briefly occluded, causally interacting objects and false beliefs. It’s not just initial proactive gaze ...
... they also succeed when you measure relatively long looking times in a violation of expectations task.
For example,
20 seconds!!! Why is this relevant? The early-developing process is supposed to be fast, whereas how long you look at something appears to depend on how interesting it is to you. (At least when you’re looking times are in the tens of seconds.)
On the face of it, looking times in the tens of seconds are just the wrong sort of thing to be an upshot of a fast process.

Spelke et al. (1992, p. figure 6 (part))

in Onishi & Baillargeon (2005), whose subjects were 15-month-olds, we have looking times of roughly 17 vs 27 seconds.
10 seconds!!! Why is this relevant? The early-developing process is supposed to be relatively automatic, whereas how long you look at something appears to depend on how interesting it is to you. (At least when you’re looking times are in the tens of seconds.) On the face of it, looking times in the tens of seconds are just the wrong sort of thing to be an upshot of a relatively automatic process.
Consider also the standard view about why infants look longer. They had an expectation about how the protagonist would act. They recognise that the expectation is not met. And now they are wondering whether they have observed the scene correctly and what they might have missed. This reflection on the violation of an expectation is very much not the sort of thing that an early-developing, relatively automatic process is supposed to enable.

Q2

How could fast processes
influence looking duration and pupil dilation?

(And why do they not influence verbal responses or manual search?)

early-developing,
automatic process

???

looking duration
(v-of-e or dishabituation)

[first question] How does the early-developing, automatic belief-tracking process measurably influence looking duration?

early-developing,
automatic process

? ? ?

verbal response,
manual search, ...

[second question] And why does the early-developing, automatic belief-tracking process not measurably influence verbal responses?% \footnote{ There are many related questions raised by a dual-process theory of development. These include: \begin{enumerate} \item What, if anything, ensures that distinct processes ever nonaccidentally provide matching answers? [synchronic] \item How do processes which are relatively unchanging over development influence the emergence, in development, of knowledge of simple facts about particular things? [diachronic] \end{enumerate} }