Thursday 5 July 2012

Bjorn Brembs Behavioral Freedom and Decision-Making in Flies: Evolutionary precursor of "Free Will"?


Abstract: The collaborative actions of chance and necessity make up the foundation of evolutionary success: the deterministic rules of selection act upon the stochastic genetic variation to bring about adaptive change. Genetics studies both the variability of genomes and the almost faithful transmission of genetic information from generation to generation. The same concerted action of chance and necessity underlies bacterial chemotaxis: Escherichia coli uses straight runs and random tumbles to orient in odor plumes. In both instances, we understand both the mechanisms underlying the generation of variability and those of the the deterministic components. The behavior of organisms with nervous systems also employs this powerful combination when at first different behaviors are tried out in a new situation until the desired goal is achieved. Subsequent encounters with the same situation then lead to the successful behavior increasingly quickly. While we understand the deterministic selection processes ('reinforcement') leading to the reliable production of the behavior comparatively well, we know next to nothing about how the behavioral variability is generated that provides the substrate for these selection processes to act upon. Mutation, sexual recombination, jumping genes or horizontal gene transfer are crucial not only for evolution to take place, these fundamentally stochastic processes also make evolution principally unpredictable. Analogously, the processes by which brains generate variable and sometimes genuinely new behaviors are crucial for brains to generate adaptive behavioral choice and make brains principally unpredictable. It is this unpredictability which forms the evolutionary basis for behavioral freedom, a candidate for the evolutionary precursor to what we today call 'free will' in humans.

      Brembs 2010 Towards a scientific concept of free will as a biological trait: spontaneous actions and decision-making in invertebrates. Proc Roy Soc Bhttp://rspb.royalsocietypublishing.org/content/early/2010/12/14/rspb.2010.2325
      Brembs (2008) The Importance of Being Active J. Neurogenetics
      Grobstein, Variability in Brain Function and Behavior
      Miller, Protean Primates: The Evolution of Adaptive Unpredictability
      Raichle, Two views of brain function
      Doyle, Free will: it's a normal biological property, not a gift or a mystery.

Comments invited

49 comments:

  1. Distinguish (1) the question of determinism (is everything that happens just part of the cause-effect chain determined by the Big-Bang) vs indeterminism (quantum mechanics) from (2) the question of unpredictability (as in statistical mechanics), from (3) the question of voluntary action and choice (when I do something because I feel like it, was my feeling really a cause?).

    ReplyDelete
  2. But is there a cause to non linear behavior? Wouldn't it be a determined stochastic behavior?

    ReplyDelete
  3. It seems clear that flies do choices, but can we relate these choices to intentions? How can we link choices to intentions in general?

    ReplyDelete
    Replies
    1. Choice and intention are closely linked. I believe that when there is a choice, the is an intention (the intention would be 'what you want to do' and choices will help you reach your goal). i don' t know if flies do choices. They certainly react to operant conditioning.

      Delete
    2. without access to their subjective experience, I don't think we can say they have an intention.

      Delete
    3. However, I think it is reasonable to say that a fly intends to avoid the noxious stimuli

      Delete
    4. In this basic form, you would make intention and conditioning equivalent. Given this, a fly (or any organism, biological or not) that avoids noxious stimuli would have intentions.

      Delete
    5. I agree with Laurence. Also, if a robot avoids a stimulus, would we that this robot intended to avoid it? Probably not, it seems detecting and reacting to certain stimuli do not require 'feeling', simply 'doing'. Does the fact that it is a noxious stimulus change things?

      Izabo Deschênes

      Delete
  4. I assume the stochastic behaviour was taken to be the trademark of free will, but it's not clear to me that that should be the case. Either linear or non-linear behaviour could be interpreted as 'free will'.
    More importantly, both could also be argued to be no demonstration of free will at all. A toss of the dice is random but there is nothing to do with free will; reflex action is determined but there is no free will either.

    ReplyDelete
    Replies
    1. I agree. It is not clear to me what the relation between stochastic behavior and free will is. Free will is not a capacity to act randomly in certain circumstances. If your actions and intentions (in certain circusmtances) are determined solely by something like a random number generator, then you really are not in control of what your doing. You're not choosing anything. Free will must involve minimally the capacity to respond to your own desires and your beliefs.

      Delete
    2. That is correct, stochasticity alone doesn't do it: mutation without selection isn't evolution. Stochastic processes without deterministic processes to make us of them, do not constitute behavioral freedom.

      Delete
  5. @Muszynski: Indeed, random alone is not free, deterministic alone is also not free. But a system that contains both can be free. That's the main point I tried to get across: if you only look at one, freedom goes away. Without selection, evolution doesn't happen, without mutation, evolution doesn't happen. You need both or the phenomenon disappears.

    ReplyDelete
    Replies
    1. I'm having trouble understanding how a system that contains both linear and non-linear behaviours suggests freedom more than a purely linear system or a purely non-linear system.

      You also mentioned criticality, which seems to imply that the linear and non-linear components need to have a special relationship to suggest freedom of the system. Could you elaborate on criticality and what you think it implies?

      Delete
    2. 1) Nonlinear systems have the weird property of being able to produce, under certain parameter conditions, linear output (despite the fact that the system generating the output works according to nonlinear rules). This needs to be understood as separate from adding stochasticity to such systems.

      2) Here comes an even more weird property: if you tune nonlinear systems to be mathematically unstable (i.e., such that we would see an increasing curve in the S-MAP procedure) these systems can become indeterminate even though the nonlinear rules of the system are completely deterministic. In the 1980s this is what people called deterministic chaos.

      3) So far, everything above can be generated exclusively with deterministic components, that's fascinating in and of itself. However, when you add stochastic components to the system (such as noise), it becomes principally indeterminate, as the nonlinearities in the system can amplify the low-level noise that is present in the system.

      It sounds to me that the first part of your comment was getting to my point 3), while the second part was geared towards 1) and 2).

      If you have a nonlinear system with noise, and it is mathematically unstable, then you have highly variable output where the presence of noise is important, but how much noise you have is fairly irrelevant. If you have nonlinear system that is mathematically stable, than its output is less variable and this variability dependent on the amount of noise in the system. In the first case, the S-MAP curve would be inclining, in the second case not, even though the variability may potentially seem similar to the human eye.

      Criticality refers to systems that are hovering exactly at the edge of being mathematically unstable and can thus combine properties of both being able to generate large amounts of variability and to also clamp it down and there are only small changes to the system required to transition between these two states.

      So far, we have almost zero evidence that this is what happens in fly brains. The fly data are compatible with this idea, but so are probably a million other explanations. We are currently working on getting at the biology of these processes to make sure.

      Delete
  6. I think that there is something very important to that observation that behavior occurs outside the stimulus-response box. Be it as a result of non-linear behavior or free will or whatever. Those behaviors lead to new input, it helps to explore the world around us and we do not just sit there and wait till some trigger comes along to provoke a pre-formulated behavior. To interprete this as a pre-cursor of free will is really challenging, but i like it a lot.
    Also, that difference between world vs self learning is really interesting and i would like to talk to you Björn about it a bit more. I see some links to the question i have.. :)

    ReplyDelete
    Replies
    1. Similarly intrigued by world/self learning!

      Björn, Patrick Haggard mentioned yesterday that the has a hunch that learning to control external objects (i.e. world-learning) and learning to control our own behaviour (i.e. self-learning) probably develop in that order: first we learn about the world and then we extrapolate to ourselves. I wonder if the rutabaga-type AC appears earlier in development than PKC?

      Delete
    2. My bets go the other way around. ;) first self and the body p, then external. But still i assume that agency could be devided into different agency types: external vs internal, something i missed in several works of Haggard's lab. But would appreciate any discussion on that, if you like.. :)

      Delete
    3. Check out this paper:
      http://www.nature.com/nature/journal/v456/n7220/full/nature07590.html
      (Open Access)
      The authors claim (and I'm not competent to comment on that) that their model is the best to model the nervous system of the Urbilaterian. As you can tell from the paper, their organization is behavior first, then evaluation of the outcome: the animals first move randomly and as soon as light comes, it inhibits ongoing behavior such that they move towards the light.

      Thus, in the light of this paper, I'd hypothesize that self-learning is evolutionarily older than world-learning.

      The order is reversed, though, if you ask what happens first during the experiment in an animal like our flies: there, what you see is that during the experiment, world-learning is quick and actually slows down self-learning such that we can detect world-learning coming on after already 4 minutes, whereas this experiment has to go on for 12 more minutes before we can detect self-learning. I had to cut these results out for time reasons.

      Delete
  7. Indeed - which is why it is crucial that we found a nonlinear signature, because we know that nonlinear systems which work according to fully deterministic rules (i.e., no explicitly stochastic components) still behave indeterministically (e.g. multi-body problems). Real-world nonlinear systems (i.e. those that incorporate explicitly stochastic components) are principally indeterminate.

    ReplyDelete
  8. I'm available until I have to leave for my flight tonight, which departs at 11pm.

    ReplyDelete
  9. I am intrigued by fly "personality". In particular, the fact that after relyably learning about the world (e.g. turn right = heat) the flies once in a while would turn right as if to test their learned assumptions of the world. Is it possible to predict this behaviour in flies? For example, the rate of learning about the world (right = heat) is a predictor of the later behaviour of testing the constancy of the world ( right ? heat )?

    ReplyDelete
    Replies
    1. Excellent question and empirically answerable. We haven't done that and my subjective impression is that you will find different classes of individuals in which you will find all possible correlations. We have the data if you want to have a look :-)

      Delete
    2. Neat! I will try to catch you before you leave.

      Delete
  10. As a philosopher I think it might be useful to recall that freedom of will and freedom of action are two interdependent, but separate issues. If I am (physically) free to go left OR right I have (some degree of) freedom of action. However, this does entail I also have freedom of will. If the outcome of my choices (willings) is determined or partially constrained by external manipulation or chance - rather than stemming from my self made intention to act - I am not the ultimate source of my choices and thus I have not free will. For example, robots have freedom of action (say, move right or left), but lack genuine freedom of will (since we have programmed them to do something, and THAT is what will determine their choices - even just by increasing the probability of one over the other).

    Control over alternatives seems thus fundamental to attribute free will to someone. It is within this "garden of forking path" model of control that we can say: we have free will if we might have chosen othervise (rather than DO otherwise - since that is freedom of action).

    This is just a side note, but I think it is important to keep in mind in order to avoid misinterpreting the definitions Björn gave at the beginning of his very interesting talk.

    ReplyDelete
    Replies
    1. "If the outcome of my choices (willings) is determined or partially constrained by external manipulation"
      If this is a condition for free will, then I don't understand why anybody ever discussed this concept. I don't have wings, so I may will to fly as much as possible, this will is constrained by the external manipulation of not having wings. I may be misunderstanding your argument, as I'm not a philosopher and have only little philosophical training,but this sounds as counter-productive as the notion that freedom cannot exist as determinism isn't free and neither is stochasticity.

      It seems to me that a notion of absolute, unconstrained freedom is so artificial and so easily refuted, that I find it hard to understand why anybody is seriously proposing it. Freedom must always be constrained, only the universe (and even there we're not sure) is not affected by external forces.

      Thus, the way I understand the data, a biological understanding of our abilities to choose must always consider the question not "if" we have free will, but rather "how much". That may disappoint some, as it indeed is quite constraint, but this is still a far cry from being predictable automatons as which animals and sometimes humans are often portrayed.

      Delete
    2. 1) BB:"I don't have wings, so I may will to fly as much as possible, this will is constrained by the external manipulation of not having wings." My reply: In this case, I would say you are the ultimate source of your willing to fly (hence you DO have free will), but your freedom of action is constrained by your physical constitution, i.e., not having wings (hence you DO NOT have freedom of action).

      2) There are compatibilist arguments arguing that free will is compatiple with determinism. So it is not necessary that freedom of will is constrained by determinism.

      3) I agree with you that a notion of absolute, unconstrained freedom is ill conceived (at least for your purposes). Freedom of action is basically constrained by the physical laws of nature (gravity, and so on), and freedom of will is similarly constrained: physically (by our limited cognitive capacities), and psychologically (we are influenced in decision-making by a multitude of social, cultural, environmental, and other factors).

      The point I wanted to make was that freedom of will needs that whatever willing one has, she is the ultimate source (she is in voluntary control) of that willing (i.e., she REALLY could have chosen otherwise). I don't see how we can pretend to understand "how much" free will one has, since I don't see how we could understand whether it is the ultimate source of its choices.

      Delete
    3. Ah, thanks for clarifying that (somewhat, lol). I think the data we have suggest that flies REALLY could chose otherwise (i.e., the sum of all extrinsic causes is not sufficient to explain/predict the choice), but until we know how the neurons are doing it, we cannot have a very high level of certainty.

      Delete
    4. I am intrigued by the notion that our sense of free will could stem from the chaotic behavior in neuronal systems, and the special mathematical features of non-linear systems. I can’t help the feeling that there is something fundamentally right about this intuition: freedom only makes sense within a framework of constraints. However, it seems to necessarily lead to the idea that philosophical free will, as mental causation, is radically inadequate. Indeed, if neuronal decision making and criticality is all there is to human volition, then the processes that engage action are purely physical; therefore, mental events would have to be *effects* of physical processes, and never causes.
      This is a massive headache for anyone who has dabbled in Action Theory, but it seems to be correct.

      Delete
    5. Separating 'mental' and physical is dualism (see my first few slides). 'Mental' is just another word for 'brain processes' - and of course brain processes can cause any action (as well as other brain processes). By replacing 'mental' with 'brain processes' your problem dissolves.

      Delete
    6. BB: "replacing 'mental' with 'brain processes' your problem dissolves"

      Wish it were that simple. Besides, the problem is not mental vs. physical but felt vs. done. And replacing "felt" by "brian processes" doesn't tell you how or why organisms feel rather than just do.

      Delete
    7. I agree with Dr. Brembs here, and believe that Maxwell's statement is a fallacy. Of course human volition derives from neural activity (physical events). However, the supposition that the physical properties producing one mental state are always effects and can therefore never influence subsequent mental states is nowhere supported.

      Delete
  11. I am fascinated by Dr. Brembs' and others' findings that dysfunctions in specific synaptic-plasticity molecules which are expressed brain-wide presumably can lead to deficits in very particular types of learning like language/song acquisition and self/world learning. I want to commend Dr. Brembs for showing us what flies can be trained to do and for illustrating how behaviour-genetics-neuroscience can all be studied in the same experimental framework!

    ReplyDelete
  12. I commend Brembs on committing to an operational definition of free-will. It might not be consistent with our views of free will, but it at least gives us a starting place to explore from.

    ReplyDelete
  13. As Bjorn Brembs pointed out nicely, all brainy organisms have inputs, those inputs are generated by the brain and it results in outputs as what we call behavior meaning that all kind of behavior is a stimulus-response. In other words, behavior needs a stimulus to occur, which makes great sense since behavior can be seen as the result of brain activity only and that this activity has to be triggered at some point by a starting stimulus.

    He also pointed out that in an ambiguous situation organisms had to make a decision to reach their "species-related goals" coping with the stimuli present in the environment. However, my first question/comment is the following.
    Does it really exist a "behavioral freedom" when you have to behave to fulfil your primary basic needs for survival and reproduction? Because, I can explain that in more details if needed in another comment but, behavior only serves organisms to ensure survival and reproduction (and according to the psychobiological appraoch of behavior, even any human behavior would have evolved because at some point, it serves survival and/or reproduction). So even if you "make decision" to behave, this behavior will always be motivated by a the aim to increase the probability to maximise survival and reproduction. Considering that, is there really a "freedom" in our behavior or, do we always "decide" to behave according to the best estimation of our brain to increase our fitness?

    My second question/comment is about the first experiment with the flies in a completely stimulus-free environment (free of ambiguous situations).
    All the following is mere guessing but... Even in that stimulus-free environmnent, is that really a completely free stimulus-free environnent? Because in this experiment, only external stimuli are taken into account. But what about internal stimuli? because as long as an organism is alive, brain needs to be active. To be active neurons in the brain need to be activated, but to be activated, neurons need a starting stimulus. Evolution works to create adaptive behavioral patterns that has to be encoded somewhere in the genes. So even in the absence of stimuli, an organism has to have a behavioral strategy to cope with an ambiguous-stimulus-free environment. In the absence of external stimuli, organism has thus to produce internal stimuli to guide their behavior. In other words, developping a sereotypic behavior in the absence of stimuli would itself be adaptive because in that kind of situation you always need to find strategies that will lead you to fulfill your needs for survival and reproduction. Memory might have a really important role in that process.
    To summarize, I think behavior in the absence of stimuli is never mere randomness, not to say it has nothing to do with randomness. On the contrary, behavioral strategies involved in decision making has to be somehow encoded and guided by memory contents.

    ReplyDelete
    Replies
    1. Actually, I said the opposite: all brains are action-outcome evaluation, not stimulus-response. Stimulus response is a highly derived trait and mainly a laboratory artifact.

      The question about fitness is excellent and tricky, because you ask about ultimate and not proximate causations (that's a distinction in biology you can lookup if you're not familiar with the nomenclature).
      Indeed, increasing your own fitness is a main driver for your behavior,which is precisely why we try to make the situation as artificial as possible. In such a situation, the general situation may elicit some general strategy, but that strategy doesn't explain at what specific timepoint neurons should fire to make the fly go 'left' or 'right'. For example, the strategy "I have to get our of here" doesn't explain why the animal is flying straight now, left a millisecond later and hundred milliseconds later makes a right turn: during all of those behaviors, the strategy remained the same.

      Thanks for bringing this up, I think this was a very important aspect that I didn't have time to go into.

      Delete
    2. What is an 'internal stimulus' and how does it make the fly go left at one time point and right at the next?
      BTW, the behavior we see in the flies are precisely the opposite of 'stereotypical', they're highly variable and you are right, they are not 'mere randomness': they are in-between 'stereotypical' and 'mere randomness', that was the most important part of my talk: just like evolution or bacterial phototaxis isn't 'steretypical' or 'random', spontaneous behavior always contains both components.

      Delete
    3. I have been pondering a similar question about the "stimulus-free" nature of the fly's "sensory deprivation tank" (as one of the morning's commentators put it). The fly is literally glued to an apparatus that is preventing it from flying away. It is, however, awake and free to flap its wings, move its antennae and a host of other things. Thus it appears to me that the purported "spontaneous" behaviour consisting of repeated attempts to fly to the right and left is likely to be escape behaviour. This fits well with what Pauline was saying about behaving to maximize fitness - in this context we can probably say that all animals are "programmed" to seek freedom of movement in order to best search for food, a mate, etc. Thus the "random" left and right movements are expressions of this programming. And thank you Bjorn for pointing out that there is nevertheless a certain freedom within the programmed strategy with respect to the choice of direction at any given moment in time (as a homo sapien who has evolved via natural selection may I just say: whew, what a relief!) In terms of stimulus, I also don't know what a purely internal stimulus would consist of in a fly, since I can't imagine it ruminating over its memories or anything like that. But I do suspect that the fly is receiving some sort of stimuli from its sensory system indicating to it that it's attempts to fly away are in fact failing, and that these signals may in fact be the cause of it's repeated attempts to move left and right. This still doesn't explain why it "chooses" left or right in the way that it does, but it does make the claim that the fly is acting in the "absence of any stimuli" seem like somewhat of an overstatement.

      Delete
    4. What Sarah pointed out is quite similar to what I reply on Facebook to Bjorn. Here is my reply :

      Thanks everyone for all theses comments.

      But I keep thinking that randomness might be, exactly as I think free will is, mainly a mere ilusion. Because I can't see how we can have a non-dualistic way of thinking and include some sort of randomness? it has to have something that triggers the action. Neuron firing may appear random at some point (and I insist on the word "appear") but how do you explain the apparition of the stimulus that fires it in the brain?

      However I totally agree on a certain kind of randomness in terms of probability in the sense that in the absence of relevant stimuli, it is quite adaptive to distribute equally the probability of developping a certain behavior rather than another (wich seems to me to be merely the basis of trial-error mechanisms which is the most efficient way of learning how to react in physiological time to the environment), but I think this "randomness" of the bahavior is itself encoded in the genes as what seems to me to be an adaptation and that the first random behavior (such as turn left or right) that will be produced may at least have a probability of occuring slighty higher considering the effect on the organism in terms of benefits of all the stimuli the brain recorded until the moment the behavior is made. I mean that the brain might do its own probality of a certain behavior to be effective and always choose that with the highest propability of improving the fitness.
      In addition, That might be questionable but I think that as long as the organism already have a "program" encoded in its genes to produce the "randomness" of the behavior, this randomness looses its intrisic "randomnessibility"...

      Delete
    5. Clearly, in intact animals, complete absence of stimuli is impossible. The important thing is that nothing in the environment is triggering each behavior as the stimulus situation is constant.
      Perhaps more convincingly, one finds similar spontaneous behavior even in isolated nervous systems, see here for an example video from leech (but I've worked in isolated Aplysia ganglia as well):
      http://www.youtube.com/watch?v=-DmH3NxnUvM

      Clearly, there are no sensory organs left, so any 'internal stimulus' is generated by the neurons in the dish, which would mean the term 'stimulus' refers to anything that goes on in neurons and then this term would be rather useless.

      If you think randomness is an illusion you give the impression of a member of the determinist religion: many adherents, no evidence.

      In general, it's important to distinguish between randomness/stochasticity and the random-like behavior we have been measuring. Randomness is just a component that interacts with deterministic, nonlinear neuronal circuits to accomplish random-like behavior. That distinction is critical for understanding the behavior I described.

      Delete
    6. COMMENTS COPIED AND PASTED FROM THE CORRESPONDING POST ON FACEBOOK :

      ANDY NDK :
      "It is still some sort of randomness i think, but as he pointed out not complete randomness in mathematical terms, but randomness set up by a nervous system, where we will be able to find some sort of principals. And i agree: also in that stimulus free environment the animal seeks to fulfill its biological needs and try to "survive". But the quasi-random behavior helps here: it hopefully leads to new stimuli, hopefully stimuli relevant to the survival. And i think that was the principal idea of Björn's talk: that possibility to quasi-random behavior should be implemented in the animals nervous system, this will increase hopefully fitness. And to take this even a bit further: this could maybe an explanation why humans are very succesful evolutionary speaking as we might exhibit the biggest potential for quasi-random, stimulus-independent behavior. Therefore his idea that this trait could be precursor of what we call free will is very tempting and i assume provocative to most people."

      MICHAEL SHALDEN :
      "Randomness exists in behavior, owing to noise, but it does not buy much. For ethics, free will, responsibility, "Chance is as relentless as necessity" (Simon Blackburn). I think the neurobiology of decision-making does help ground an argument about moral responsibility at least and maybe FW too. It helps to focus not on the low level neural events or the noise but on the policies that the brain applies. A good example of a policy is the tradeoff between speed and accuracy. I like this example because we know something about the neural mechanisms (I told you about this in my lecture) and because it mediates the type of distinction we care about when we hold someone ethically responsible for a good or bad decision. If you want to read more, check out http://​www.ncbi.nlm.nih.gov/pmc/​articles/PMC3332233/"

      BJORN BREMBS :
      "Actually, I said the opposite: all brains are action-outcome evaluation, not stimulus-response. Stimulus response is a highly derived trait and mainly a laboratory artifact.

      The question about fitness is excellent and tricky, because you ask about ultimate and not proximate causations (that's a distinction in biology you can lookup if you're not familiar with the nomenclature).
      Indeed, increasing your own fitness is a main driver for your behavior,which is precisely why we try to make the situation as artificial as possible. In such a situation, the general situation may elicit some general strategy, but that strategy doesn't explain at what specific timepoint neurons should fire to make the fly go 'left' or 'right'. For example, the strategy "I have to get our of here" doesn't explain why the animal is flying straight now, left a millisecond later and hundred milliseconds later makes a right turn: during all of those behaviors, the strategy remained the same.

      Thanks for bringing this up, I think this was a very important aspect that I didn't have time to go into.

      What is an 'internal stimulus' and how does it make the fly go left at one time point and right at the next?
      BTW, the behavior we see in the flies are precisely the opposite of 'stereotypical', they're highly variable and you are right, they are not 'mere randomness': they are in-between 'stereotypical' and 'mere randomness', that was the most important part of my talk: just like evolution or bacterial phototaxis isn't 'steretypical' or 'random', spontaneous behavior always contains both components."

      Delete
    7. PAULINE CLAUDE :
      "Thanks everyone for all theses comments.

      But I keep thinking that randomness might be, exactly as I think free will is, mainly a mere ilusion. Because I can't see how we can have a non-dualistic way of thinking and include some sort of randomness? it has to have something that triggers the action. Neuron firing may appear random at some point (and I insist on the word "appear") but how do you explain the apparition of the stimulus that fires it in the brain?

      However I totally agree on a certain kind of randomness in terms of probability in the sense that in the absence of relevant stimuli, it is quite adaptive to distribute equally the probability of developping a certain behavior rather than another (wich seems to me to be merely the basis of trial-error mechanisms which is the most efficient way of learning how to react in physiological time to the environment), but I think this "randomness" of the bahavior is itself encoded in the genes as what seems to me to be an adaptation and that the first random behavior (such as turn left or right) that will be produced may at least have a probability of occuring slighty higher considering the effect on the organism in terms of benefits of all the stimuli the brain recorded until the moment the behavior is made. I mean that the brain might do its own probality of a certain behavior to be effective and always choose that with the highest propability of improving the fitness.
      In addition, That might be questionable but I think that as long as the organism already have a "program" encoded in its genes to produce the "randomness" of the behavior, this randomness looses its intrisic "randomnessibility"..."

      ANDY NDK :
      "I am not sure if this randomness is really encoded in the genes as you say. I would rather say there might be a genetic predisposition for an animal to tend more to "randomness" than another animal. I imagine it a bit like this: lets say u have 2 pools of neurons encoding the decision to turn left or right, so if in any of both 5 neurons fire, the animal turns left or right. First of all think how many combinations of neuroms firing together you can have which could lead to a left or right movement (and think if you scale up that pool to thousands or tenthousands of neurons) Lets say now in absence of a stimulus the decision is up to a probalistic process, that 5 neurons fire together just by chance. That process could be regarded more or less as a truly random process. I think what could be genetically determined in that process is the threshold of the decision: in one animal 4 neurons might enough, in another 6. Or it could be that the individual neurons have a different rate of random spikes, so in one animal those neurons fire more often and by this increase the chance that 5 neurons fire together. An animal with a threshold of 4 would exhibit more often a spontaneous / random movement, the other not so often. However, the genes would never encode the outcome of that process as such, only how certain features like threshold. therefore i would still call the outcome of that process a random event."

      PAULINE CLAUDE :
      "That's the common comment I have when I say "genetically encoded". What is genetically encoded is surely not the outcome (in that case it is not the fact that the flies turn left or right), what is encoded though is the process that allows the organism to test the different possibilities to solve a problem (the absence of sitmulus is itself a problem) by giving an initial probability of success of 0.5 for any given possibility that the organism can meet.

      Actually, I'm quite convinced that when it turns to behavior, all what is encoded in the genes are "abilities to" that can be activated, deactivated or modulated considering the different stimuli received from the environment."

      Delete
    8. BJORN BREMBS :
      "Clearly, in intact animals, complete absence of stimuli is impossible. The important thing is that nothing in the environment is triggering each behavior as the stimulus situation is constant.
      Perhaps more convincingly, one finds similar spontaneous behavior even in isolated nervous systems, see here for an example video from leech (but I've worked in isolated Aplysia ganglia as well):
      http://www.youtube.com/watch?v=-DmH3NxnUvM

      In these dissected nervous systems, there are no sensory organs left, so any 'internal stimulus' is generated by the neurons in the dish, which would mean the term 'stimulus' refers to anything that goes on in neurons and then this term would be rather useless.

      If you think randomness is an illusion you give the impression of a member of the determinist religion: many adherents, no evidence.

      In general, it's important to distinguish between randomness/stochasticity and the random-like behavior we have been measuring. Randomness is just a component that interacts with deterministic, nonlinear neuronal circuits to accomplish random-like behavior. That distinction is critical for understanding the behavior I described."

      Delete
  14. So flies are 'free' but not 'actually free'? What is the difference between 'actually free' and 'free'? We are our brain. If the (neuronal!) causes for our actions are in our brain, we are agents and if these actions are not fully determined we are free. That degree of freedom might not be satisfying to everyone - tough luck, dualism isn't an option, we have to live with the cards we are dealt. For me personally, the fact that 'free will' can be provided by an entirely materialistic brain is cause to celebrate, not to resign. Brains are creative, unpredictable and chaotic while at the same time able to allow us to survive in a Newtonian macrocosm. What else could we possibly want? Our brains provide us with every possible trait necessary for very action from genocide to artistic masterpieces and science. Our brains manage to do this without the need for magic and allowing us to consciously experience all this to boot. What could possibly be missing from that account?

    Unless I misunderstand you, the requirement of 'actual freedom' sounds to me like a wish for superpowers. There are comics for that. In reality, we already have a brain that works in such intricate ways that we have trouble letting go of magic in order to explain it.

    ReplyDelete
  15. By characterizing brain activity as creative, unpredictable and chaotic, I think you are buying in to some of the magic :)
    I think the brain (and it's functions) is unbelievably organized and predictable - we are just not yet intelligent enough to be able to predict it

    ReplyDelete
  16. Your last sentence contradicts the data, as it suggests indeterminate systems controlling behavior. If we were able to predict fly behavior fully, then this would be akin to a comeback of determinism that physicists have tried and failed for 80 years.
    Like I said in the talk, determinism is like a religion: plenty of people believe in it, despite all the evidence to the contrary...

    ReplyDelete
  17. Xavier Dery ‏@XavierDery

    Brembs paraphrased: "Unless the brain is a deterministic bubble in a probabilistic universe, we have to forget determinism" love it #TuringC

    11:20 AM - 5 Jul 12 via Twicca Twitter App

    ReplyDelete
  18. Xavier Dery ‏@XavierDery

    Brembs: "Being boring and predictable leads to a detrimental mate choice" : remember to act on that if you are single!! #TuringC

    11:38 AM - 5 Jul 12 via Twicca Twitter app

    ReplyDelete
  19. Seeing those drosophilia being tested for free will with a laser beam made me wonder, very naively: are there ethical guidelines for the treatment of drosophilia in research? Or is it accepted that they are not animals, therefore have no feelings, and as such can be manipulated however the experimenter deems informative?

    ReplyDelete
    Replies
    1. I'm not sure whether there are defined ethical guidelines for Drosophila here at McGill, but I can tell you that PIs working with Drosophila don't have to write AUPs. PIs working with Aplysia, though, do have to write use protocols, despite the fact that they have nervous systems that are much less complex than those of the common housefly (this is much to Dr. Sossin's chagrin, I'm sure!).

      Delete