Who’s In Charge? Free Will and the Science of the Brain by Michael S. Gazzaniga
On the evening of October 10th 1769, in one of his typically curt dismissals of a philosophical problem, Dr Johnson silenced Boswell, who wanted to talk about fate and free will, by exclaiming: “Sir…we know our will is free, and there’s an end on’t.”
That anecdote is related by Anthony Gottlieb, who in a recent essay reminds us that the history of those who would deny free will is long and varied. The latest, and certainly not the last, attack is from neurologists and academics who think they understand neurology. These folks hold with strict determinism, a position which argues that we should “not hold people accountable for their actions or anti-social behavior.” After all, their genes—no, I mean brains—made them do “it”, whatever it is. David Stove called “selfish genes,” the penultimate theory of how mysterious entities secretly rule our lives, “genetic Calvinism.” Following this, we can label brain science Neurological Predestination.
Nevertheless, there are still some of us who cannot but agree with Dr Johnson, who believe that observation trumps theory. Michael Gazzaniga is one of the rare neurologists who isn’t ready to abandon the idea that we have minds and responsibility. He made his fame by thinking up slicing through the corpus callosum to cure epilepsy patients. He gives the full history of this procedure and its consequences, which is fascinating. But we needn’t dwell on that here.
Think of the corpus callosum as a telephone trunk or bundle of wires which carries signals from the left to the right halves of the brain and back again. Notice this is “signals” and not “all signals.” Chopping through this trunk cured, or mostly cured, the epileptics who consented to have their heads cracked open. But that didn’t mean the separated brain didn’t find ways of talking to its severed self.
That’s because it’s strictly wrong to think of the brain as being an isolated organ that sits encased in a skull. It’s also the spinal cord and the nerves that snake out of it and all throughout the body. It’s even, if you want to think of it this way, the whole body, because those nerves that terminate out to pinkies and toes communicate with the tissue surrounding them.
The start of the idea of the brain as a deterministic machine comes from observations that prove that the mind constructs a view of reality as a function of various stimula. Now, there are many of examples in the book, and doubtless the interested reader knows plenty, so I won’t repeat all of them. Besides, Gazzaniga provides the best and clearest example I’ve seen. Touch your nose. You feel the pressure on your nose and finger simultaneously, yet the signal has to travel a lot farther from your finger to your brain than from your nose. That you “feel” the two at once means your mind is piecing together the two events separated in time. What becomes aware—your mind—is different than a machine itself, which first knows of the nose and only much later of the finger.
Deterministic Gedanken Experiment
Gazzaniga helpfully reminds us that the brain is a “vastly parallel and distributed system, each with a gazillion decision-making points and centers of integration.” A gazillion is a lot, so here is my example. Suppose instead there is just one point, one neuron which accepts inputs and has one output. These inputs can be as many as you like (but finite); they are simple chemical channels the body uses to talk to the neuron. The output behaves similarly.
The neuron takes the different inputs and, via some internal logic, combines them into an output. In other words, the output is a (non-linear) function of the input. This is how statisticians and computer scientists model neurons, anyway. That makes this neuron fully deterministic. If we know the inputs exactly, we can predict the output exactly, as long as we know the form of the function. And if we don’t know the form, given enough input and output data, we can do a pretty good job guessing it.
Now, since this is a real neuron we’re imagining, things won’t always work as they would in a mathematical model. The same inputs might produce different outcomes at different times, possibly because of quantum mechanical effects, or maybe because of other vagaries in the chemistry, or say because of changes in the structure of the neuron (nothing contingent lasts forever). And it could also be that we can’t measure the inputs to sufficient precision. If the algorithm governing the neuron is chaotic, it could appear that the “same” inputs generate different outputs, even widely different outputs. But that flexibility in the output doesn’t mean the workings of the neuron aren’t deterministic: they still are; we just can’t figure out the form of the determinism.
So far, we are in a fully determined world, even though we might not be able to measure it perfectly. Now add in neuron number two and wire it to the first. What changes? The number of connections, for one, and our difficulty in measuring. But that still does not change the deterministic nature of this two-neuron “brain.”
You can see where this is going: add in as many neurons as you want, and even add in neurons of different “ability” or function. It becomes an impossible mess to track and measure practically, of course, but a “massively parallel” deterministic system is just as deterministic as a non-massive, single neuron. To repeat: our ability to measure, model, or predict the form of this determinism is irrelevant to our understanding that the system is deterministic. Why, the brain in the sense we have been describing is just like a big computer.
Logic Intrudes
But our minds are not computers. There are two simple ways to prove this. Here’s the first. Take out the cheapest pocket calculator you can find and use it to calculate digits of π. It will be a fairly slow process, and one which will not fool anybody into thinking the calculator is conscious. Now transfer your algorithm to a “super computer”, perhaps a “massively parallel” one. The only thing that will change is that the digits of π will come faster. More calculations per second does not a mind make. That super computer will also not know what it is doing. This realization was what dashed the hopes of early “artificial intelligence” researchers.
The second example was given to us by Searle1. You’re in a room with a rulebook and box of cards on which Chinese characters are written. You can’t speak or read Chinese. There are two holes in the walls, one for input, one for output. When a string of Chinese symbols comes as input, you look up in the rulebook (written in English) that says, “with this string of input, send that output.” And so you send that output. The input is then switched to output and vice versa. You then wait for new input via the old output, and send the new output via the old input. Get it?
Chinese speaking (reading) people outside the room will be under the impression you/the room speak (read) Chinese, when as stated you haven’t a clue. You as “the room,” or computer, pass the Turning test with respect to those outside the room. But just because you have followed a rulebook, i.e. that you have behaved deterministically, it does not mean, and it is not true, that you understand Chinese as a human mind understands Chinese. Understanding requires more than determinism, it requires a mind.
Gazzaniga does not hold with the strict determinism of the “causal chain gang” which so infatuates many of his colleagues. We have minds and free will. Even though some of what we feel is certainly determined for us—unless your name is Shadrach, Meshach, or Abednego, you can’t will the feeling of cold by being tossed into a furnace—we still have the ability to make choices. Perhaps choices more limited than has previously been allowed, but choices just the same.
So what are his candidates for a mind? He does not argue for dualism. He has a short go at chaos theory and quantum mechanics, but never develops those themes beyond stating that measurement is difficult. He does not appear to be aware of Roger Penrose’s arguments2 about the inability of quantum mechanics to explain the mind. And we saw above that difficulty of measurement is not a proof against (nor for) determinism.
Next time: measurement, emergence, and why people hold with determinism.
—————————————————————————
1Best short treatment (I left out many details) outside Searle himself is Ed Feser’s Philosophy of Mind pp. 155-159. If you’re looking for just one book by Searle, get Philosophy in a New Century: Selected Essays. This book is also highly, very highly, recommended for statisticians with an interest in philosophy. Next in line is his The Rediscovery of the Mind.
2See his Shadows of the Mind: A Search for the Missing Science of Consciousness, or even better his The Emperor’s New Mind: Concerning Computers, Minds, and the Laws of Physics. This latter book also contains the best description of the EPR paradox I’ve ever seen.
I’ll be away from the computer until late Thursday.
This ‘we have no free will’ or ‘free will is just an illusion’ is just another ham-handed ploy at diffusing personal accountability for one’s actions. Its a sophisticated modern-day equivalent of the old joke, “the devil made me do it.” But some actually try to really believe this.
Such people can be fun to mess with. At work, steal & eat their lunch, or parking spot, etc. etc. Then claim its biology or whatever & you cannot be held accountable as you had no willful choice in the event…. When they complain about something, want a product refund, etc. keep pointing out that those responsible for, accountable for, their problem really aren’t as they didn’t have any free will in the events leading up. Etc. What emerges is a picture of someone that doesn’t want to be personally held accountable…but not so for anyone else.
I think we all agree that a scientific definition of free will would require a means of measuring it. However, IMHO, arguing from determinism is not the way to go. Consider the difficulties presented by this video. For a definition only in terms of determinism, are these two pieces of fabric exercising free will? (I am not saying they are. But my defense would include more than just an argument that the Navier-Stokes equations are deterministic.)
I’d like to preface this critique with my own personal opinion: I believe in free will, but I do not claim to have evidence for or against it.
The Emporers New Mind is a great read– very educational; the conclusion is weak though and, as with Penroses final argument, I don’t follow your argument regarding the Chinese Room.
It’s been over a decade, but IIRC in the Emporers New Mind Penrose makes a fatal assumption– that ‘self’/intelligence/free will requires the sort of ‘sub-quantum precision’ that seemingly affects the brain. I’m not convinced. A metal arm is just as capable of lifting a cup as a meat arm, despite missing 99.999% of the underlying ‘sub-quantum’ magic. Why can’t a non-meat mind get by the same way?
The Chinese Room is a limited turning test designed to measure the perception of intelligence. All we can say of free will in the Chinese Room is that the participants are willfully engaged in a game of card swaping.
Lets play a different game that REALLY tests for the presense of free will. All you need to do is answer the following questions:
– Does a waking man have free will?
– Does a sleeping man have free will?
– Does a man in a coma have free will?
– Does a dead man have free will?
If you’re willing to leave the realm of man:
– Does a Chimpanze have free will?
– Does a Parrot have free will?
– How about a fly?
– A jellyfish?
– A Plankton?
– An amoeba?
– The N1H1 virus?
– A single grain of quartz?
And now the bizzare:
– Does a human, an entity made of a collection of living cells, have free will?
– Does an ant colony, as an entity, have free will?
– Does a city, as an entity, have free will?
If you are convinced that free will exists, for logical reasons alone, then I would very much like to know where you see and don’t see it.
Here’s a test of “free will” — pick a region in which all laws & controls (police, etc.) are put in abeyance…and observe what happens. Places such as where riots occur serve as a good proxy, or the real thing. Race riots, in Los Angeles not too long ago (early 1990s) for example, are illustrative: If strictly biological processes were determinitive, why would false/misleading sensationalized inflammatory press releases of a trial(s) outcome provoke such outrage & the subsequent lawlessness among some & not others?
And, would those that espouse such views of “free will” also endorse the abolition of the entire criminal justice system…after all, if everything is biologically determined these are really superflous, right?
Many participants & non-participants in riots, when interviewed, explained a variety of reasons for their behavior varying from opportunism weighed against risks of getting caught to those who intervened to help, selflessly & at personal risk & cost, out of altruistic noble values … & other behaviors encompssed everything in between those extremes.
Given the similarity of basic biological “design” of individuals it would seem reasonable to conclude that if biology dictates behavior (no “free will”) then the resulting behavior relative to a given environment would be relatively consistent….but what’s observed is radically diverse.
So diverse that predicting what any individual will do is essentially impossible…so…even if “free will” (however defined) were truly dependent on biology, given the diversity & unpredictability of outcomes arising from a given stimulus/stimuli, the unpredictable nature of resulting behavior effectively makes a distinction, as far as humans are concerned, between determinism & free will impossible.
Ken: And, would those that espouse such views of “free will†also endorse the abolition of the entire criminal justice system…after all, if everything is biologically determined these are really superfluous, right?
Well, no, not really. Wouldn’t the proponents of such a view also argue that the criminal justice system is just as biologically determined too. After all, that’s what “everything” means. This response is adapted from an old Muslim answer to an argument about Fate. The executioner has no more choice than the murderer. All is predetermined.
Mind you, it doesn’t look that way to me either.
Will,
I will take a crack at answering your questions.
My opinion is that free will is a property of sentient (self aware) beings. I have seen some people try to argue that a worm is sentient because it can sense it’s environment. This is in my oppinion a missunderstanding of the term. That I am aware, the term was first used in science fiction to separate people aliens from animal aliens. The term sentient in this context does not refer to awareness of ones environment, but to awareness of one’s self.
A simple test for sentience would be the mirror test. Put the test subject in front of a reflection of itself and does it recognize that the reflection is itself or does it act as if the reflection is another entity.
Does a waking man have free will? Yes
Does a sleeping man have free will? Yes
Does a man in a coma have free will? Maybe, No one knows what is going on inside the mind of a coma patient. The fact that the choices can not be comunicated to the outside world does not eliminate free will. If the coma patient is still cabable of thinking then yes.
Does a dead man have free will? Taking this to refer to the physical body, no.
Does a chimpanze have free will? Probably. Most of the apes will pass the mirror test at least as adults, though some even if exposed to a mirror from birth will not pass untill they reach a minimum age. My understanding is that human infants will pass from birth (or as soon as the eyes are open. The lesser primates will generally not pass the mirror test.
Does a parrot have free will? Maybe
How about a fly? No
A jellyfish? No
A plankton? No
An amoeba? No
A virus? No
A grain of quartz? No, inanimate objects in general cannot have free will as they have no will.
Does a human, an entity made of a collection of living cells, have free will? Yes
Does an ant colony, as an entity, have free will? Does a city, as an entity, have free will? No and no. Personally I reject the premis that a collection of entities is an entity in it’s own right.
But our minds are not computers. There are two simple ways to prove this.
If you mean computer = calculator, then no, people aren’t all that good at it. A bit pointless as It doesn’t prove that a computer as a general device can’t think or understand.
Understanding Chinese: how exactly would you go about proving someone/anything understands Chinese? Some variation of the Turing test? Explain the difference between a human mind understanding Chinese and a non-human’s non-understanding. Do you have to know in advance if a black box that apparently understands Chinese actually contains a Chinese? Isn’t that a bit like predicting tomorrow’s weather the day after tomorrow?
Understanding requires more than determinism, it requires a mind.
You know this how? Is it your definition? If so, please define “mind”. Something that “understands”? Full circle: define “understands”.
Ken,
If strictly biological processes were determinitive, why would false/misleading sensationalized inflammatory press releases of a trial(s) outcome provoke such outrage & the subsequent lawlessness among some & not others?
Are automobiles deterministic (determinitive?) ? If so why doesn’t filling the tank result in the same performance among all cars?
would those that espouse such views of “free will†also endorse the abolition of the entire criminal justice system…after all, if everything is biologically determined these are really superflous, right?
Huh? What does that have to do with anything? Free will has to exist because you think the world would be worse without it? Really?
Many participants & non-participants in riots, when interviewed, explained a variety of reasons for their behavior varying from opportunism weighed against risks of getting caught to those who intervened to help, selflessly & at personal risk & cost, out of altruistic noble values … & other behaviors encompassed everything in between those extremes.
Sounds like support for decision-making based on cost/benefit analysis.
So diverse that predicting what any individual will do is essentially impossible
Which is also true of the weather. Practically impossible is not the same thing theoretically impossible. Knowing every input is what makes prediction practically impossible.
I agree that the argument is endless as it can never be put to a test. We can never say when we have found free will. No different than arguing the number of angels that can dance on the head of a pin, that here is no monster under the bed, etc.
Matt: thanks for humoring me. You might enjoy the philosophers network. They have some neat online logic tests relating to this sort of thing.
You answered that a dead person has no free will. What happens to a person who is clinically dead, and then reanimated? They had no free will, and then they suddenly have it again. This would imply that you believe free will is an active process rather than a specific state (the dead and living are made of the same stuff after all).
I’m curious why you are willing to assume a primate in a vegetative state might have free will, but not an ant colony that is actively shaping its environment.
The mirror test can be failed by a self conscious person. A blind person, for example. Certain neurological disorders might also make it impossible for someone to pass many different variants of the test.
I’m not up-to-date on how Searle’s Chinese Room Argument has developed in the decades since he first presented it, but there’s an obvious error in the argument he made then. In his article in Scientific American, he somewhat formalized the argument, and one element of this was to claim that a computer program only has “syntax,” not “semantics.” He claimed that the output of the Chinese Room could be interpreted as stock quotes or anything else, rather than remarks in Chinese.
That’s obviously false – this jumped out at me while I was reading the article.
For example, try to play chess with Microsoft Word. That is, try to interpret what it does as the chess moves of an opponent. The probability that this is possible is extremely tiny, because it requires a *consistent* interpretation as chess moves. The same goes for the Chinese Room output. Try interpreting it as the text of “Romeo and Juliet” in English, or even in Chinese, if the conversation has nothing to do with “Romeo and Juliet.” You don’t find a consistent interpretation that does the trick.
But in case such an interpretation were possible, the probability would be overwhelming that there was a heretofore unknown relationship between Word and chess-playing, or between the Chinese conversation and “Romeo and Juliet.” It would probably mean that our understanding of reality was so far wrong that we could be judged insane.
So I say that a computer program does contain semantics, but this does not mean that it is conscious. If Searle thinks that semantics implies consciousness, he thinks a non sequitur.
Strict determinism is NOT ‘a position which argues that we should “not hold people accountable for their actions or anti-social behavior.‒ Indeed determinism can be used to *justify* holding people to account as it admits the possibility that our response to past behaviour is capable of influencing future behaviour (and by the same argument, at least some level of determinism might be *required* for such justification)
Will,
“You answered that a dead person has no free will. What happens to a person who is clinically dead, and then reanimated? They had no free will, and then they suddenly have it again. This would imply that you believe free will is an active process rather than a specific state (the dead and living are made of the same stuff after all).”
Personally, I would be inclined towards thinking of free will as a process. However, in considering free will as a state, free will is an energy state, not a mater state. While a living body and a dead body are comprised of the same mater, their energy states are drastically different.
“I’m curious why you are willing to assume a primate in a vegetative state might have free will, but not an ant colony that is actively shaping its environment.”
There are different sorts of vegetative states. If there is still brain activity, we have no idea what is going on inside. Such a person could be caught in some kind of hallucination/dream state with the brain cut off from the rest of the body. In such a state some form of free will could still exist. My point in saying maybe to this case is that with current technology, we have no way to make a difinitive test.
A collection of multi-cellular organisms is not equivilent to a single multi-cellular organism. If a single ant does not have free-will then the colony as a whole can not. Offer evidence that some form of self aware hive mind exists in a colony of ants and I will re-consider.
“The mirror test can be failed by a self conscious person. A blind person, for example. Certain neurological disorders might also make it impossible for someone to pass many different variants of the test.”
This interpretation is not correct. As I posit it, the mirror test is only failed if a entity is aware of the reflection and acts as if the reflection is a separate entity. A blind person would make for a null or inconclusive result, not a fail. Also, I have not claimed that the mirror test is dispositive by itself. In the case of neurological disorders, further evidence would be needed for a dispositve result.
Also, I use the term reflection here loosely. It would be interesting if we could construct a sonic mirror to test creatures such as dolphins that percieve their evironment primarily through echo-location.
As Alan points out, and I have commented before, accountability and responsibility require determinism. You can only conclude that a damaged brake pipe is responsible for an car accident or someone’s poor impulse-control is responsible for aggression, if you accept that the car and person behave deterministically. Without a causal link there is no justification for punishment or repair.
Lets look at the two “proofs” that minds are not computers:
1) A supercomputer can calculate pi without knowing what it means
I don’t see how that is meant to prove the claim, but in any case I can get a schoolchild to calculate pi with pencil and paper without knowing what pi means. By Briggs’ same “logic” that means the schoolchild does not have a mind.
2) The Chinese room rule-reader doesn’t understand Chinese, but the room system does.
All this shows is that a small part doesn’t have the capabilities of the whole (the “rulebook” needs to be much more complex than the rule-reader). Equally a chess-player’s neuron or a circuit in a chess-playing computer doesn’t understand chess. This has no relevance to determinism. Also note that a real brain or computer needs to be dynamically reconfigurable with memory and adaptable heuristic features to have a credible conversation in Chinese, so “a rulebook” is clearly a misleading analogy for the strong-AI system required.
Toby: I think you’re wrong on both counts.
1. The argument you’re criticizing does not say that if an entity can calculate pi without knowing what it means, this shows that it does not have a mind. Rather, it says that if the entity can calculate pi AND does not have a mind, no endowment of additional computing power with which it will calculate pi can be evidence that, with the additional power, it has a mind. This is a straw-man argument; no one claims that the calculation of pi, by itself, if enough power were used, would be evidence of a mind. But it is not the argument you are criticizing.
2. You are imposing restrictions on both the rule book and the man in the room which the the CRA (Chinese Room Argument) does not impose:
a. There is a man in the Chinese Room who is obeying the rule book, and we presume that he has dynamic memory and so forth – else how could be obey it?
b. Some computer programs use heuristics, so there’s no reason why they could not be encoded in the rule book. The CRA assumes that they *are* encoded in it.
Pingback: William M. Briggs, Statistician » There’s An End On’t! Gazzaniga’s Who’s In Charge? Reviewed: Part II
Smoking Frog,
1) The straw man here is not Toby’s but Briggs’. Noone has ever suggested that being able to calculate Pi was evidence of “having a mind”. In fact, the question of whether there is some higher level of computational activity which might be equivalent to having a mind would not even be addressed by a much stronger argument – showing e.g. that *no* numerical calculation is such evidence. It may be instructive to note that the French word “ordinateur” creates less confusion in this context as it does not identify the management of complex networks of linked information that a “computer” does with mere calculation.
2) The ‘Chinese Room’ scenario may serve some paedagogical purpose as a conversation opener but as an “argument” it is pathetic. I agree with your criticism of the reference to “semantics” in Searle’s SciAm article but I also think Toby’s main point is valid. The important part is not the quibble about what we mean by “rule book” but the fact that the operator’s understanding (or not) of Chinese is irrelevant to the question of whether or not the entire room-system can be said to “understand”. There may be good reasons for denying that the room has understanding (though I’ve never seen any that don’t amount to just declaring it) but the operator is just another man of straw.
I don’t see how a deterministic reality equates to “lack of accountability.” First, there is the argument that people determined to do wrong must be locked away for the common good. Second, you can say that people who do bad things simply “are” bad people.
The machine may be complex, but I think we are in a world of determinism, meaning whatever happens couldn’t have happened any other way. I am with “compatibilism.”
Aristotle wrote that anger can be explained either as a boiling of blood around the heart or by a desire for revenge. Take the second as a version of free will and the first as an early version of determinism. (When the highest tech was hydraulic systems, men interpreted mind in hydraulic terms – hence, the humours, etc. Later, machines were the highest tech and mind was regarded as a mechanical contrivance. Now computers are the ne plus ultra, and we are using new metaphors. We should not forget that they are metaphors.)
What is clear (or ought to be clear) is that anger is intelligible in both modes. As a medical condition, something like the “boiling blood” metaphor would be applied. As a legal matter, the “desire for revenge.” It’s not clear why some folks insist on all one thing. It’s like arguing whether a basketball is rubber or a sphere.
Alan Cooper: Perhaps I wasn’t clear.
1. I agree with you that Briggs’ argument is a straw-man argument, but I say that Toby mistook it for another argument and criticized that argument.
2. Toby says that “The Chinese room rule-reader doesn’t understand Chinese, but the room system does.” is alleged to be a proof that “minds are not computers.” He tries to show that it is not a proof that minds are not computers. Well, it is not alleged to be a proof that minds are not computers. To say that the room system understands Chinese is to say, or at least to suggest, that minds *are* computers.
This makes it hard to follow his subsequent argument, but as far as I could see, he treats the man in the room as if he were not part of the “room system,” which is incorrect, regardless of what he’s trying to prove or disprove. You’re right that whether the man understands Chinese is irrelevant, but I’m not aware that anyone claims otherwise.