Theory of Mind, Working Memory and Inhibition

The ability to understand other minds (aka Theory of Mind, or ToM) seems to develop rather abruptly between the ages of three and four. Some suggest that this reflects a true "conceptual shift" in the mental capacity of children, and point to the fact that several tasks thought to measure ToM show improvement around the same time.

However, other researchers view ToM as resulting from a more domain-general change in cognitive skills - in particular, those related to executive function (EF; e.g., inhibition, updating, working memory, etc). Advocates of this perspective point to variance shared by ToM and a wide-variety of EF tasks, which remains even after removing variance due to age and verbal ability. This researchers suggest that ToM is either made possible by improved EF, or that the capacity for ToM is simply made easier to express by improved EF.

Support for the latter idea - that competence on ToM tasks is revealed but not directly caused by improvements in EF - comes from several studies which have found that ToM performance is very sensitive to task demands. For example, performance on false-belief tasks can be manipulated by changing the inhibitory demands of the task (for example, using a "magic pointing stick" in a deceptive pointing task seems to alleviate the inhibitory demands by physically separating children from the item they are acting upon). Others have found that ToM is sensitive to differences in memory demands - for example, having children draw and later view a picture of their previous beliefs improved performance on a task where they then were asked to recall this earlier belief that was actually inaccurate.

So, is it inhibition or memory (or both) that allows children to demonstrate their competence on Theory of Mind tasks? A 2005 article by Hala, Hug and Henderson explores this question by administering a series of executive function and ToM tasks to three and four year-old children.

In the past, the relationship between working memory and ToM performance has been somewhat unclear, often disappearing altogether after the effects of age and verbal ability (VA) are partialled out. Hala et al. suggest that the few WM measures that still correlate with ToM after removing covariation due to age and VA are characterized by their strong inhibitory demands. For example, backwards but not forwards digit span correlates with ToM after partialling age and VA, and putatively requires "inhibition" in that one must not perform the more natureal forward task. [Although the authors do not mention this, the updating and maintenance requirements of backwards digit span are higher as well, so this does not clearly implicate inhibition alone].

According to Hala et al., the case for inhibition's relationship to ToM performance is more clear-cut, in that tasks with both inhibitory and memory demands are consistently related to ToM, whereas inhibitory tasks along are not so strongly related. For example, tests of inhibition that involve only delay were not as strongly related to ToM as those that also involved competing or conflicting responses.

In the current investigation, Hala et al. attempted to use a more "pure" measure of memory demands, although they acknowledge that almost all memory tasks may involve some inhibition. Casual readers may wish to skip the methodological details that follow in italics.

Here are the tasks they used:

• Inhibition tasks: gift delay , in which kids had to not peek at a gift that was being wrapped. Time until peeking (if the kids peeked at all) was the dependent variable.
  
• Working memory tasks: Stroop-control task and the six-box scramble task were used, in which kids simply had to respond with "day" or "night" to two arbitrary visual stimuli, and in which kids had to search for six stickers hidden inside a set of six differently-patterned boxes whose locations were "scrambled" after each search.
• Combined Working memory and Inhibition tasks: Day-night Stroop and Luria's tapping task were used, in which subjects had to respond "day" to a picture of a night-time scene (and vice versa), and in which subjects had to tap twice if the experimenter tapped once, (and vice versa), respectively.
• False-Belief tasks (thought to measure ToM)
  
• "Uncued tasks"
  
• Unexpected location task
• Deceptive location task
• Unexpected contents task
• "Cued tasks"
• In these tasks, kids were asked to draw a picture of the initial "expected" location or object contents. Kids were also told that this picture would help the protagonist later remember where the object was (which is different wording than typically used in FB tasks). There was a cued version of each of the uncued false-belief tasks listed above.
• Control measures:
• PPVT (a test of verbal ability)
  

The results showed, surprisingly, that performance was not significantly better on the "cued" false-belief tasks than on the "uncued" false belief tasks. If you believe that the cueing procedure eased the memory demands, then this might suggest that memory alone is not important for false-belief performance or ToM.

Interestingly, the false-belief tasks involving locations did not correlate with those involving contents, although each of the location tasks correlated with each other, as did each of the contents tasks. Performance was numerically higher on the content tasks than on the location and deception tasks. This may indicate that content false-belief tasks may pose less memory or inhibitory demands than false-location tasks.

Even after partialling out age and verbal ability, the false-belief contents task was significantly related to the WM EF tasks and the Combined-WM&IC EF tasks, whereas the false-belief location tasks were related only to the Combined-WM&IC tasks. Only the Combined-WM&IC tasks significantly predicted false-belief performance.

Based on these results, the authors conclude that ToM, as measured by false-belief tasks, is related to a combination of working memory and inhibition, but not to working memory alone. The critical pieces of evidence that support this conclusion are

1) only the tasks involving conflict and memory demands (i.e., day/night Stroop and the tapping task) significantly predicted false-belief performance, and

2) manipulations designed to reduce the WM demands of false-belief tasks did not significantly improve children's performance on these tasks.

However, there are a few reasons to think that this conclusion is premature:

Regarding the first point above, these tasks are extraordinarily process impure: the six-box scramble task is said to measure memory alone, but seems to clearly involve a strong inhibitory component and yet did not correlate with false-belief performance after partialling out the effects of age and verbal ability; likewise, the control Stroop task was very easy for all the kids in the experiment, and is a very limited and unusual test of working memory to begin with.

Regarding the second point above, clearly these manipulations may not have substantially reduced the memory demands of the tasks. Hala et al. themselves note that drawing may have also distracted children from the task at hand. Manipulations involving emphasis of the differences between the expected and unexpected locations or contents may be a stronger test of the hypothesis that false-belief performance is improved by reduction of memory demand. This would explain why false contents tasks are somewhat easier for all age groups.

Related Posts:
What Matters for Theory of Mind?