Earlier today I discussed some of my problems with Alfred Mele’s 2014 book Free: Why Science Hasn’t Disproved Free Will. Like Dan Dennett, I agree that the book is somewhat tainted by being funded by and associated with a foundation (Templeton) that undoubtedly loves Mele’s ideas, but I’m not at all accusing Mele of writing the kind of stuff he knows that Templeton wants. Templeton skews academic discourse by selectively funding those scientists who find out stuff it wants, not by getting the scientists it funds to produce what the Foundation wants.
But onward and upward. There was one statement in Mele’s book that sounded weird to me, and it was about Drosophila, so I looked up the reference. On page 81 of Mele’s book we read this:
So is there hard evidence of deep openness? [JAC: Mele means that decisions are open regardless of what happened before them; in other words, true libertarian free will]. Some scientists say they’ve found something like it in fruit flies (Brembs 2011). If a fruit fly turned to the left a moment ago and time were wound backward a couple of moments, it might turn to the right in the rerun. There’s no evidence that fruit flies have free will. Instead, there’s evidence that something that people regard as necessary for free will—behavior-producing processes that aren’t deterministic—is present in fruit flies. And if such processes are present in them, they might be part of our evolutionary heritage; they might be present in us too.
What the hell? What evidence is there that at a given time a fly has behavior-producing processes that aren’t deterministic? (I assume Mele is not invoking quantum phenomena here, for, as he recognized earlier, such phenomena can’t really be part of “agency”.) So I looked up the Brembs paper in Proc. Roy. Soc. B, which you can see for free by clicking on the screenshot below (pdf here, reference at bottom of page).

Unfortunately for Mele, Brembs’s paper says nothing of the sort. The paper is largely about variability in animal behavior, and how varying your behavior can be adaptive, even in situations that are superficially similar to ones you’ve encountered before. And yes, animals do vary their behavior in seemingly unpredictable ways to avoid predators, find food, and so on, but that unpredictability says absolutely nothing about the absence of physical determinism. This is the kind of “evidence” that Brembs adduces (and yes, he adduces it in support of free will):
For instance, isolated leech nervous systems chose either a swimming motor programme or a crawling motor programme to an invariant electrical stimulus [78–80]. Every time the stimulus is applied, a set of neurons in the leech ganglia goes through a so far poorly understood process of decision-making to arrive either at a swimming or at a crawling behaviour. The stimulus situation could not be more perfectly controlled than in an isolated nervous system, excluding any possible spurious stimuli reaching sensory receptors unnoticed by the experimenter. In fact, even hypothetical ‘internal stimuli’, generated somehow by the animal must in this case be coming from the nervous system itself, rendering the concept of ‘stimulus’ in this respect rather useless. Yet, under these ‘carefully controlled experimental circumstances, the animal behaves as it damned well pleases’ (Harvard Law of Animal Behaviour).
. . . A great invertebrate example of the sort of Protean behaviour [31,32] selected for by these trade-offs is yet another escape behaviour, that of cockroaches. The cerci of these insects have evolved to detect minute air movements. Once perceived, these air movements trigger an escape response in the cockroach away from the side where the movement was detected. However, which angle with respect to the air movement is taken by the animal cannot be predicted precisely, because this component of the response is highly variable [33]. Therefore, in contrast to the three examples above, it is impossible for a predator to predict the trajectory of the escaping animal.
Note to Brembs (and Mele): “poorly understood” does not mean “determinism not at work”. Jebus, that’s an elementary mistake!
Does this unpredictability of behavior mean that it’s not controlled by physical processes—that it’s not determined by the laws of physics? Hardly. All it shows is that neither we nor other animals can predict exactly what a leech or a cockroach can do in circumstances that seem similar. But there could be plenty of determinism acting here. For instance, there could be an internal program that says, “If you swam recently, crawl now, or vice versa.” Or there could be all kinds of different things going on at the neuronal level that we don’t understand, yet are determinative in the animal’s behavior. This kind of stuff is simply the conflation of indeterminism with lack of predictability: a common error in discussing free will.
But on to the flies. Here Brembs is referring to studies of tethered fruit flies put in a chamber with a moving grate in front of them, to which they will respond as if they were flying and the grate was “moving” because they were flying by it. Their turning behavior, however, is variable:
For instance, in the study of the temporal dynamics of turning behaviours in tethered flies referenced above [63], one situation recorded fly behaviour in constant stimulus conditions, i.e. nothing in the exquisitely controlled environment of the animals changed while the turning movements were recorded. Yet, the flies kept producing turning movements throughout the experiment as if there had been stimuli in their environment. Indeed, the temporal structure in these movements was qualitatively the same compared with when there were stimuli to be perceived. This observation is only one of many demonstrating the endogenous character of behavioural variability. Even though there was nothing in the environment prompting the animals to change their behaviour, they kept initiating turning manoeuvres in all directions. Clearly, each of these manoeuvres was a self-initiated, spontaneous action and not a response to some triggering, external stimulus.
Yes, sometimes the flies’ turns were unpredictable, and they turned in various directions. Does this mean that their turns were not determined by physical processes, that, as Mele described the results, the turns “weren’t deterministic”? Not at all! All it says is that in a similar situation, “exquisitely controlled”, flies can do different things. Perhaps they have a program that, like Chuck Yeager in his crashing plane, says “Try X! Then try Y! Then try Z!”. They may have variable behavior, even in a constant environment, because they’re seeking some outcome that the experimenter doesn’t know, or simply because they have a neuronal program that evolved to produce variable results in nature and is also activated in this study. And it may not depend on any external stimuli, but so what?
In his paper Brembs talks a bit about chaos theory, which shows that in some processes tiny differences in initial conditions can be amplified into big differences in results. That might mean that, in such an experiment, there are imperceptible differences at different times in the “exquisitely controlled environment” that result in different directions of the turn. It’s another common error to think that chaos theory somehow negates determinism. Chaos theory is a deterministic theory.
You don’t have to be a genius to see that there are purely deterministic explanations for Brembs’ results. I believe, then, that in his book Mele has distorted Brembs’s results in such a way that they look like they favor his “ambitious free will” view. That’s why I argue that Mele’s book is tendentious.
Two last points. In an attempt to save libertarian free will, Brembs himself talks about the kind of unpredictability that would give humans moral responsibility—a system that has a way to generate nonrandom results. And so he highlights a two-stage model of free will. To wit (my emphasis):
. . . it is precisely the freedom from the chains of causality that most scholars see as a crucial prerequisite for free will. Importantly, this freedom is a necessary but not a sufficient component of free will. In order for this freedom to have any bearing on moral responsibility and culpability in humans, more than mere randomness is required. Surely, no one would hold a person responsible for any harm done by the random convulsions during an epileptic seizure. Probably because of such considerations, two-stage models of free will have been proposed already many decades ago, first by James [94], later also by Henri Poincaré, Arthur Holly Compton, Karl Popper, Henry Margenau, Daniel Dennett, Robert Kane, John Martin Fisher, Alfred Mele, Stephen Kosslyn, Bob Doyle and Martin Heisenberg (cited, reviewed and discussed in [7]), as well as Koch [9]: one stage generates behavioural options and the other one decides which of those actions will be initiated. Put simply, the first stage is ‘free’ and the second stage is ‘willed’. This implies that not all chance events in the brain must manifest themselves immediately in behaviour. Some may be eliminated by deterministic ‘selection’ processes before they can exert any effects. Analogous to mutation and selection in evolution, the biological process underlying free will can be conceptualized as a creative, spontaneous, indeterministic process followed by an adequately determined process, selecting from the options generated by the first process.
But this doesn’t take the determinism out of free will—not unless the “options” generated in Stage 1 are generated by quantum-mechanical processes. Mutations might involve such processes, but brain functions? We have no idea. And even if the “options” (presumably the range of things you are capable of doing and know you’re capable of doing) are generated by quantum mechanics, that is still physical determinism.
Second, Brembs trots out the old canard that we should believe in free will because without it people would behave badly—a conclusion often drawn from the reference below:
Finally, there may be a societal value in retaining free will as a valid concept, since encouraging a belief in determinism increases cheating [103]. I agree with the criticism that retention of the term may not be ideal, but in the absence of more suitable terms, free will; remains the best option.
The reference here is to the well known Vohs and Schooler 2008 paper (citation below), which has failed to be replicated at least twice since it was published (see here). Further, an extensive analysis last year showed no evidence that belief in free will correlated with either prosocial or antisocial behavior. It’s time to stop citing the Vohs and Schooler paper as evidence that we should believe in free will as a kind of social lubricant. (We shouldn’t, of course, accept what isn’t true just because it makes us behave better, but that’s another issue.)
There’s more in Mele’s book that is dubious, but I think I’m done for the time being. All I can say is that people will go to great lengths to take the determinism out of free will. Compatibilists do it by saying that you can have your determinism and free will at the same time, while mushy writers like Mele and Brembs do it by positing some kind of indeterministic process that affects one’s choices but doesn’t involve quantum mechanics. In the end, if you think that determinism simply means that your behaviors must obey the laws of physics on all levels, then you can’t have agent-determined you-could-have-done-otherwise free will.
________
Brembs, B. 2011. Towards a scientific concept of free will as a biological trait: spontaneous actions and decision-making in invertebrates. Proceedings of the Royal Society B: Biological Sciences 278:930-939.
Vohs K. D. and J. W. Schooler J. W.. 2008. The value of believing in free will: encouraging a belief in determinism increases cheating. Psychol. Sci. 19: 49–54.doi:10.1111/j.1467-9280.2008.02045.x (doi:10.1111/j.1467-9280.2008.02045.x)