Developing Intelligence

Imagination allows us to escape our current time, place, or perspective in favor of an alternative context, whether that may be fanciful or mundane. So imagination is a mechanism for specifying and maintaining a context that differs from our more immediate and stimulus-driven environment (at least, that is what I mean by "imagination"). According to Buckner & Carroll, this kind of "self-projection" from one context to another is the essential function that underpins the involvement of the prefrontal & medial temporal lobe (MTL) circuit in a variety of tasks, including those requiring retrieval from long-term memory, planning, theory of mind, and possibly even navigation. This remarkable similarity across tasks is illustrated in the image at right, taken from the Buckner & Carroll paper.

Although somewhat vague, this idea is not as crazy as it might sound. Buckner & Carroll remind us that others have espoused similar views. For example, they review arguments that the prefrontal cortex can 'transpose the effective reference point from self to other, from here to there, and from now to then." Other recent work has emphasized the role of this circuit in "mental time travel."

Buckner & Carroll also emphasize the role of the medial temporal lobe in imagination. For example, the neuropsychological patient HM is famously amnestic but, less famously, is also impaired at planning and making predictions about his own future. This is also true of other patients, including D.B., who seems selectively impaired at autobiographical predictions. Such patients can still imagine generalities about the future, but are impaired when asked to imagine specific aspects of their own future.

Recent neuroimaging of these tasks demonstrates that similar circuits are activated by detailed planning, theory of mind, and episodic recall, and that these same circuits are also part of the "default network" of brain regions which tend to become active when subjects are not given any instructions at all. That is, "undirected" tasks (such as staring at a blank screen) are accompanied by a "highly stereotypical pattern" of brain activity that overlaps strongly with those involved in tasks requiring imagination. This default network may reflect an adaptive tendency for humans to mentally simulate alternative events or contexts, so as to prepare for those contingencies. (Or, it may simply be what people do when you ask them to stare at a blank screen while in the scanner...)

Nonetheless, this perspective is very compatible with the perspective reviewed yesterday, in which Burgess & Hitch argue that understanding the varying roles of "context" is important for revealing the relationship between long and short-term memory. This compatibility can be illustrated in the way both theories bear on recent data by Conway et al., who revealed that left prefrontal regions show more activity when subjects are thinking of an imagined autobiographical event than a real autobiographical event, and that occipito-temporal regions show the opposite pattern (caveat: some might doubt that large slow negative-going shifts in scalp electrical potential reflect increased neural activity, as argued by Conway et al.).

Why should real autobiographical events elicit activity in sensory regions? According to Burgess & Hitch, items in long-term memory (in which these real autobiographical events reside) are related to one another directly. Various aspects of the experience may therefore prime other aspects of the experience, leading to a "chain reaction" of recall. According to Bruckner & Caroll's perspective, it makes perfect sense that this "chain reaction" of recall would manifest as greater activity in sensory regions.

And why might imagined autobiographical events elicit more activity in frontal regions? First of all, there is no long-term memory for a newly imagined event; assuming then that the imagined event exists in short-term memory, the items or events which comprise the memory will need to be temporally sequenced with respect to an imaginary "timeline." This requirement matches almost exactly with the characteristics of short-term memory reviewed by Burgess & Hitch, in which items are associated with an abstract, time-varying and multidimensional "context signal."

So despite yesterday's hesitations with regard to the Burgess & Hitch perspective, it does seem to have utility in explaining these characteristics of imagination... but there's still plenty left to explain.

For example, why does the frontal cortex seem particularly important for imagination of autobiographical events (whether real or imagined)? This question relates to some fascinating developmental evidence that the divide between reality and imagination is far thinner for children many would have expected, but that the line is more blurred between reality and imagination for autobiographical than interpersonal events. I'll start reviewing this evidence next week.