2.2 Bias and Intention

summary

Chun Siong Soon’s 2008 fMRI experiment

Subjects had the task of pressing a button with their right or left hand whenever they felt the urge. There were no “go” signals in the experiment.
During these tasks the subjects’ brains were being recorded using BOLD fMRI, an imaging technology that measures blood flow in the brain as a way to determine areas of brain activity.
The authors claimed that the brain activity they recorded could predict which button a subject would press 7 to 10 seconds before they actually pressed it, with 60% accuracy.

Itzhak Fried’s 2011 depth electrode study

When epilepsy patients cannot be well-treated with drugs, they receive brain surgery to reduce the occurrence of seizures. If they consent to participating in scientific studies, electrodes can be placed directly on their brains during surgery. These depth electrodes are similar to EEG, but have even better signals since they are beneath the skull.
The task for subjects was very similar to Libet’s study–they pressed a key on a keyboard whenever the urge arose. The subjects all had depth electrodes that recorded their brain activity during the experiment.
Like the other studies mentioned, the authors said they could predict from the recorded brain activity when a subject would press a button, 800 milliseconds before it happened, with 80% accuracy.

Reaction time studies support the notion that conscious decision-making takes place

An important question to ask in regards to all of these experiments is: how long does it take for a conscious decision to actually generate action?
There is indirect data already available to start to answer this question, in the form of reaction time studies. In these studies the subject knows what action they need to do when they receive a “go” signal. This is important because it means that they do not know when they will need to make a decision, so subconscious decision making cannot take place.
One such study from 1999 found that the mean time between the “go” signal and the beginning of muscle motion was 231 milliseconds. The time between intention and muscle motion must be even less, since the subject must first hear and process the “go” signal before generating intention and action.
Another study involved a “decide” signal. Subjects were told to choose a hand to use for a button press only once they heard a tone, and then to immediately use that hand to press the button. This study found that the mean time between hearing the tone and acting was 150 milliseconds, and there was no difference in pre-action brain activity for left- and right-hand presses.
These studies both place decision making in the same time frame as the conscious perception of decision-making in the previously described experiments. This suggests that conscious decision making does take place.

What did the Libet experiment and its successors actually detect?

Libet’s studies and the other described experiments did find some neural activity that preceded conscious decision making and could often predict future behavior.
It is possible that what was being detected is a neurological bias toward a certain behavior. This bias is not deterministic but simply increases the probability of certain behaviors. This would explain the fact that none of the studies could predict behaviors with 100% accuracy, and some were not much above chance.
This is no problem for free will because it means the brain has not definitely decided on an outcome before the subject becomes aware of it; it just nudges the subject in one direction, and they can go against this nudge.
Even the Fried study, with its 80% accuracy, does not indicate that a decision has actually been made 800 milliseconds before action. It is consistent with the idea that the brain is getting closer to making a decision at that time but has not completely done so.

The Libet Study and the Point of No Return

Even if these experiments only detect bias, it is possible that the bias can be strong enough to pass a point of no return. Basically, the subconscious bias can be so large that no real conscious decision-making happens.
To refute this there would need to be data showing that a large buildup in brain activity did not predict action.
Libet’s study had a major flaw that prevented this from being seen: brain activity was only recorded for trials when the subject moved their arm. This means there is no data to show examples of brain activity building up with no action taking place afterward.