Reason as Attentional Prosthesis

post by Gary Basin (gary-basin) · 2017-11-14T00:23:28.277Z · LW · GW · 1 comments

Contents

1 comment

[cross-posted from: http://garybasin.com/reason-as-attentional-prosthesis/]

What are we doing when thinking, or reasoning?

Often, thinking is goal-directed. There is some outcome we are trying to achieve or problem we are trying to solve. In hindsight, answers often seem obvious. “Why didn’t I think of that?” For the most interesting problems, there is no formula or system that will lead to an optimal solution. The hard part becomes framing the problem. How do we do that? I have some thoughts on the role of attention.

But first, pigeons! In an experiment, pigeons are rewarded for pecking a toy banana hanging from the ceiling of their enclosure, but only while standing on a box placed below it. They’re also rewarded, in a separate setup with no toy banana present, for moving a box to randomly marked spots on the wall of their enclosure.

And then they’re faced with a novel puzzle:

“When these pigeons were confronted with the banana out of reach, the box present but displaced from under it, and no spot was on the wall, they looked back and forth between banana and box at first, appearing “confused” to human observers. Then, within less than two minutes, they abruptlybegan to push the box in the direction of the banana, stopped when it was underneath, climbed and pecked.” (Epstein, Kirshnit, Lanza, & Rubin, 1984).

A conundrum that stresses the bird brain. Its attention is drawn to the banana, motivated by reward previously reinforced, and yet something is wrong. The pattern is not quite the same as before. Our feathered friend has yet to encounter a banana and a box at the same time where the box was not under the banana. Also, he has not been previously rewarded for moving a box except to a marked target location on the wall. To solve this problem, the pigeon seems to need to somehow suppress its default, habituated behavior. We can see it struggling to do so. Perhaps this is the pigeon-equivalent of “dropping preconceptions”, or our habitual way of seeing a situation. Once a primary reaction is suppressed, it can hold attention on other options which don’t immediately seem like answers to its problem.

Even more is demanded of the pigeon. He must figure out how to combine familiar behaviors in a new way. The pigeon experiences pleasure — anticipated reward — when he sees a box and a marked spot on the wall. Now, there is no marked spot. He also experiences pleasure when he sees a box under a banana. Now, the box is elsewhere. How to fit these pieces together to cause new behavior? Presumably, he is doing something like reasoning. He is aware of the box, an ability (affordance) to move boxes, and the good feelings associated with a box under a banana. He must be able to ignore other aspects of the scene which are not relevant. By first simplifying his experience to relevant aspects, he can attempt novel combinations. Once perceiving a candidate solution, new action is a foregone conclusion — the pigeon is quick to push the box across the cage.

Reason seems to serve as a kind of attentional prosthesis. By default, humans, and pigeons, are responding reflexively — according to habit. When reflexes aren’t good enough, we need some way to explore pieces of past experiences and recombine them in new ways. In the process, we need help navigating our attention away from what is often the most attractive stimulus. Reason functions as such a mechanism. We recall memories of situations where only certain aspects (“abstractions” or “concepts”) are attended to. For each aspect, we trigger a re-cognition — a re-experiencing — of previous experiences that are somehow related. The relation can be a similarity of stimulus, but also, crucially, similarity of outcome or response. This is also known as functional or mediated generalization. By compressing into relevant aspects, the mind seems to be more capable of chaining together previously unrelated “sub-experiences”. If a chain is conceivable to an emotionally-appealing outcome, action can follow.

1 comments

Comments sorted by top scores.