Consciousness and how it got to be that way

Tuesday, March 29, 2011

Button-Pushing Experiments Disprove Only One Fragile Model of Freewill

Experimental evidence continues accumulating which shows that human motor behavior, and conscious awareness of the decision to perform the behavior, are often preceded by neuronal activity. (Classic Libet experiment described here.) This has sometimes been taken as a disproof of free will. This reasoning is flawed. If we believe "if free will exists, then it applies to all human motor behavior, and all neuronal activity engaged with this motor behavior must be conscious". Therefore, for these experiments to be a disproof, at least two assumptions must be true: 1) the entire chain of events from deciding to act to performing the act must take place within consciousness to be called free will, and 2) this simple motor task model must be generalizeable to all decision-making and motor behavior taking place on all time scales.

To #1: it is possible that the decision is made, relative to the depolarization path of the motor act, "prior" to consciousness. If this is always the case, then epiphenomenalism is the correct description of consciousness, even though we could still have free will, because we can still affect future belief states and motor acts. Consequently it's worth asking if in these models, the "originating" neuron(s) is/are really the originating neuron. Acts of apparent free will merely passing through an unconscious pathway would be even less troublesome to the idea of free will than one originating in unconscious neurons.

Analogously, we cannot choose beliefs - we don't have free will about what we perceive and believe, because perception and belief follow necessarily from sense data and from the model of the universe that exists in our brains. When we encounter sense data, we interpret it immediately as representing something and as being causally related to other things in reality. For example, you cannot see someone smile, and DECIDE consciously whether or not to believe that they are smiling, and whether or not to believe that this means something about their emotional state. But this doesn't mean that we are prisoner to those beliefs and that we cannot influence future perceptions and beliefs. If you cross-check those beliefs, you can engage in behaviors the outcomes of which can differ in obvious
ways that then will allow you to update those beliefs.

At the very least, concluding that experimental results of this sort invalidate the idea of free will, you must carry other assumptions about the nature of consciousness and free will besides just "some objects' motions cannot be predicted from prior states, and those objects have conscious experience which in some way influences those motions". The separability of these two parts of free will - the idea of unpredictable output on the one hand and input from conscious experience on the other - is important. What are the requirements for non-conscious objects with unpredictable output? Are there conscious objects with completely predictable output? Another way of asking is whether things without nervous systems can have free will.

To #2, it's worth asking whether there are classes of decisions that are more or less subject to free will; i.e., how much can we generalize these results? Certainly there are cases where free will can be an illusion, but this falls short of disproving the possibility of free will in general. By all means experimenters should continue asking this question in more settings, although a button-push model wouldn't seem to give the nervous system a chance to shine. My suspicion is that if free will exists in any action it's in long-term executive planning.

It's worth mentioning that it seems strange very strange that our own lack of free will is not obvious to us, given that prediction of the behavior of conspecifics is evolutionarily extremely important. Why the inability to do so exactly, or why the strong appearance that we are the source of action? One speculation is game-theory oriented, that the input-output chain must necessarily be clouded by introducing randomness (which wouldn't really be free will), and our sense of free will falls out of that.

Another attack on the concept is Churchland-like, and questions whether free will isn't just a bad folk idea about human behavior (which I think many enthusiasts of these experiments which agree to). It's worth asking whether belief in something like free will is universal even among non-Western humans - do Solomon Islanders and pygmies have the same belief? - and if so, what it is that predisposes humans to believing this. Do humans with a different conception of agency have different beliefs about this (Aspergers spectrum people)? Finally, do nonhuman animals also believe in free will? I encourage inventive experimentalists to think of a way to support or discredit this hypothesis. If we know different kinds of nervous systems have the same belief we can start understanding what it is we have in common with them that makes us think this.