The cognitive impenetrability of the visuomotor system (part 1)

One thing I think I'll find eternally interesting is the degree to which my brain is doing things that I think I'm doing. That sounds painfully confused, so I'll put it another way. Dennett gave us the personal/sub-personal distinction (thanks, Dennett!). I'm consistently fascinated by how much what I thought was personal is actually sub-personal. I'm not totally sure why it fascinates me so much. I can't deny that the brain becomes more beautiful to me as I begin to "see" my brain in my behavior, and maybe that has something to do with it. There's also that great conundrum -- weakness of will. Why do we do things that, at some level, we don't want to do? Conversely, why do we fail to do things that we want to do? It would be nice to know.

Those are big questions that are perhaps only conceptually related to the question that I really want to ask: How much does explicit prior knowledge of a reaching task help us determine a strategy for that task? Not much at all, according to a few recent studies.

Here's the story. Joo-Hyun Song and Ken Nakayama (2007) gave participants two basic reaching tasks. The “easy” task consisted of pointing to a single target on a screen, and the “hard” task consisted of pointing to an odd-color target among distractor targets. These two tasks were presented in three different conditions: blocked, mixed (i.e. pseudo-randomized), and alternating. They measured reaction times for each reaching movement.




The homogenization effect

In a setup like this, you'd expect the reaction times for the blocked easy task to be fast, and those for the blocked difficult task to be relatively slower. And that's exactly what happens. But when you randomly mix the easy and hard tasks, you get something called the homogenization effect, which essentially is an attenuation of the differences in reaction times between the easy and hard tasks. People get slower on the easy tasks and faster on the harder tasks.

What's behind this homogenization effect? There are (at least) two approaches to answering this. The commonsense approach is to say that explicit prior knowledge of the upcoming trial type gives people the opportunity to optimize their strategies, and since it is missing in the mixed condition, their strategies are suboptimal (i.e. too slow for the easy and too fast for the hard task).
The other approach is to say that people build up an optimized strategy by a sort of short-term motor memory. This memory would be implicit, passively accumulated, and quick to dissipate. Thus, if a certain trial type was repeated several times, the visuomotor system would accumulate an optimized strategy for the next trial (regardless of what you thought was coming next), and that accumulation would be abolished if the trial types were mixed in a random order.

So which one is it? Explicit prior knowledge or cumulative learning? We can figure it out by pitting trial-type repetition against explicit prior knowledge. Song and Nakayama did this by including the third condition: alternating back and forth between easy and hard tasks. We know there will be a homogenization of reaction times for the mixed condition, but will the same thing happen for the alternating condition? Or will explicit knowledge of the upcoming trial type allow the participants to optimize their reaching strategy?

Here are the results.




If anything, the alternated condition looks even more homogenized than the mixed condition. This is a strong argument in favor of the cumulative learning hypothesis. They go on to show that the number of trial-type repetitions actually influences the gradual optimization of motor reaching strategies in a linear fashion until the most optimal strategy is reached. They also show that the suboptimal strategies permeating the mixed and alternated conditions result in curved trajectories toward the odd-colored target (in the hard task). It is as if people (or their visuomotor systems) are guessing and moving before they have selected the appropriate target, and correcting mid-flight if they happen to be going to the wrong target. And all of this is happening while we have the distinct impression that our explicit knowledge is going to give us an advantage. It doesn't.

This study is a first step in uncovering the degree to which the visuomotor system is cognitively impenetrable. More to come soon.


References

Song, J.-H., & Nakayama, K. (2007). Automatic adjustment of visuomotor readiness. Journal of Vision, 7(5):2, 1-9