The End of So-Called “Free Will”: “…a prediction of the effort associated with candidate movements is computed very quickly and influences decisions within 200ms after presentation of the candidate actions.”

Standard

From Abstract:  “When given a choice between actions that yield the same reward, we tend to prefer the one that requires the least effort. Recent studies have shown that humans are remarkably accurate at evaluating the effort of potential reaching actions, and can predict the subtle energetic demand caused by the non-isotropic biomechanical properties of the arm…a prediction of the effort associated with candidate movements is computed very quickly and influences decisions within 200ms after presentation of the candidate actions. Furthermore, while the MEPs measured 150ms after stimulus presentation were well correlated with the choices that subjects ultimately made, later in the trial the MEP amplitudes were primarily related to the muscular requirements of the chosen movement. This suggests that corticospinal excitability (CSE) initially reflects a competition between candidate actions, and later changes to reflect the processes of preparing to implement the winning action choice.”

Cos, I., Duqué, J. and Cisek, P. (in press) “Rapid prediction of biomechanical costs during action decisions” Journal of Neurophysiology.

While studies of decision-making have traditionally focused on the kinds of cognitive decisions that characterize human economic choices, the neural mechanisms underlying decision-making evolved long before abstract cognitive abilities.

At the time the relevant neural circuits were being established, most decisions were between concrete actions such as run left vs. right, or reach for one branch or another.  Making such “embodied decisions” entails more than just abstract representations of outcome value and includes a wide variety of sensorimotor contingencies, such as the ease of a movement or its energy requirements.  This may explain why many neurophysiological studies have consistently found correlates of decision variables within the same sensorimotor circuits that are involved in the planning and online guidance of movement.

For example, while a decision is being made:

  • neural activity in parietal and premotor regions of the oculomotor and arm movement systems encodes the potential actions in parallel
  • and is modulated by many factors relevant for a choice, including expected gain, local income, and probability.
  • Furthermore, the interactions between potential targets depend upon their spatial similarity, consistent with a competition that takes place in a sensorimotor map representing possible movement parameters.

These results can be explained by models in which potential actions compete against each other in the sensorimotor system, and this competition is biased by influences arriving from other regions, including outcome value estimates from orbitofrontal cortex, action value computation from anterior cingulate cortex, selection rules from dorsolateral prefrontal cortex, and biasing signals from the basal ganglia.  Here, we investigated how a competition between two potential reaching actions is biased by information about their kinematic and kinetic costs.  We expected that information about the relative path length would be processed very quickly because it presumably involves the fast dorsal visuo-motor stream.

In contrast, we expected that computing the more subtle biomechanical costs of the potential movements would take more time, assuming that it involves sophisticated computation through mental rehearsal or a predictive “forward model”.   Contrary to our expectations, the biasing effect of biomechanics was in fact very fast.  The effect of biomechanics was significant even if subjects were only given 200 ms to view the stimulus display prior to initiating movement.  In further contrast to our expectations, while the effect of biomechanics was equally strong at all observation intervals, the effect of relative path length became stronger between 200 and 600ms.

One explanation for this phenomenon is that the influence of path length on choices was not related to a purely spatial preference, as we initially hypothesized, but that it too was due to a preference for movements requiring less energy.  However, because path length has only a small effect on the total energetic demand of a movement, smaller than the effect of biomechanics, it may thus exert only a weak bias whose influence on the decision develops more slowly and always follows the initial specification of which muscles will produce the movement.  It is also relevant that while cells in the dorsal premotor cortex exhibit directional tuning shortly after target appearance, their modulation by path length develops gradually over 200–300 ms.

In previous studies we conducted an analysis of the contribution of biomechanics and path length to the overall energy associated with each movement and reached a similar conclusion: that the direction of movement has a major impact on energetic demand while the impact of path length is relatively small.  The behavioral results were largely corroborated by the TMS data, which confirmed that:

  • the biasing effects of biomechanics and path length were reflected in CSE as early as 150 ms after stimulus presentation…The speed with which biomechanical and geometric factors appeared to influence subject choices raises the question of what mechanisms may be responsible.
  • Previous studies have shown that neural activity in FEF discriminates pro- vs. anti-saccade instructions in 120ms.
  • In general, activity patterns across diverse regions of monkey cerebral cortex reflect a simple decision about 130 ms after cue onset. Here, we found a significant biasing effect on human CSE at around the same time, despite the fact that the biasing factors in our task would seem to require significantly more computation.

We consider two possible explanations for the speed of these effects. … Previous studies have shown that when the physical effort of candidate movements is explicitly indicated by stimulus cues, it quickly modulates neural activity in anterior cingulate cortex and, to a lesser degree, in basal ganglia…An alternative explanation for the rapidity of the biasing effect is that the brain really is able to compute biomechanical costs very quickly, and the result of this computation can quickly bias activity in the motor cortex. Indeed, if the mechanism that computes the biomechanical costs involves the same forward model that is also used in the online guidance of movement, then it would clearly have to be very fast… Although CSE at 150 ms was well correlated with the choice of the selected movement, that relationship apparently reversed at 200 ms…Thus, what we see in the time course of CSE may reflect the shifting influence of two factors:

  • First, early in the trial, we see the biasing influence of factors that determine the subject’s choice, which is made very rapidly after stimulus onset.
  • Once the decision is made, subjects can begin to prepare the muscle commands that will initiate the movement, in anticipation of the highly predictable GO signal. At this time, CSE becomes dominated by preparatory activity, which is higher during trials in which the agonist will demand a larger energy.
Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s