SAMSA: Pigeons, humans, and learning

[pErvinalia]

Anyone else can answer this as well, if they know anything about it. I'm thinking pigeons as I recall that Samsa, as a behaviourist, has talked about pigeons and their processes of learning. If pigeons don't fit this example, then substitute rats in or some other appropriate animal.

So, let's start at the end and work backwards, with a broad question about animal learning.
Q1. Are the processes of learning broadly the same between pigeons and humans? As a behaviourist, I'm expecting Samsa to say yes, and as a pretty full on materialist I'm tempted to say that yes, they are broadly the same processes.

So let's have a couple of learning scenarios
Scenario 1:
We've got a cage or box or whatever for a pigeon to go into, and that box has three buttons - a green, red and blue (substitute colours for patterns or whatever if pigeons don't see in colours, and substitute buttons for levers or whatever makes the most sense). When the green or blue button is pressed, nothing happens. When the red button is pressed, a small dosage of food is released out of a tube. Let's say we've got 50 pigeons and we perform a large enough number of experimental replications. We'd expect at the beginning for the pigeons to be randomly pressing buttons, but after enough replications we'd see a large percentage of the pigeons pressing the red button first, as they have learnt that they get food when the red button is pressed (and nothing when the others are pressed). So the pigeons have learnt.

Scenario 2:
We've got a room for a human to go into. Basically the same experiment for humans as for pigeons. Over enough replications most humans will press the red button (some might not because they don't like the food given, or are terminally stupid, or are dieting or something). I'd expect that the humans will learn quicker than the pigeons. Maybe the following question will provide a reason for that.

So, Q2 is seeking an explanation of the processes of learning shared by the two different species where there appears to be a different level of cognition going on.

To expand: What's going on in the pigeons brain while it is standing there in front of the three buttons? How does past learning manifest itself in the fact that the pigeon after enough trials is going to most likely pick the red button? Substitute for the second scenario: What's going on in the human's brain while we are standing there in front of the three buttons? This is where, IMO, there is a difference in the level of abstraction in the "mind". Human's are conscious. Pigeons aren't? I dunno, but I'm guessing they aren't.

So, back to the human. We possess a fantastic ability to hypothesise, that presumably most other animals don't possess. That is, we can decouple our thoughts/experience of the phenomenal world from creative constructs of a hypothetical reality. For example, I am sitting on my couch looking at the room around me. The TV is there, there's doors and windows and whatever. It's all pretty safe and there's no need for me to be overly worried at this point (unless I was Seth; I'd be fully locked and loaded and have nightvision gear and stun grenades at hand :p ). But I can sit here and look at this same scene and basically invent a tiger into the scene. Thanks to the human ability to decouple our thoughts from reality, I can imagine this tiger growling at me without figuratively and probably literally shitting my pants as would be the case if I couldn't tell the difference from "reality" and my imagination.

Anyway, relating this to the human scenario... We are standing there and we can recall past memories and essentially represent a hypothetical state of the world where we press either the green or blue button and no food comes out. We can then also recall and hypothesise how pressing the red button produces food. And thus we have learnt.

Now, that seems all pretty uncontroversial for the human example. But what the hell is going on in the pigeon for it to learn from past experiences? If there is no conscious recall, or at the very least no decoupling and hypothesising, how is it actually learning? There's a couple of possibilities, in my mind. 1. The process between pigeons and humans is exactly the same, and the "conscious" business going on in the human's mind is just a post hoc story we invent to explain how an event happened (i.e. us learning to pick the red button). That still leaves the question of how either of us actually 'learns'. And 2, The process is the same between the species, but the human's consciousness and decoupling abilities allow us to learn quicker than the pigeons. And there's probably a third possibility, and that would be that the process ISN'T the same between the species and there's something totally different going on, probably centred around the differences in conscious awareness.

So what's the research say on this, Samsa?

[Hermit]

Having read a bit about behaviourism almost 40 years ago, my memory is a bit vague, but as far as I can recall it pretty much ignores considerations concerning brain, mind, consciousness et cetera. Instead it focuses on action. Behaviourism is very much process driven, limiting itself to observing actions/reaction sequences. It investigates the 'how' and ignores the 'who' and 'why'. The process has been given a label: Operant conditioning. You'll find very little discussion concerning what it is that is being conditioned or what enables it to work. You may find B. F. Skinner's book, published in 1971 and titled Beyond Freedom and Dignity a useful introduction. It won't take long because It is easy to read and not much over 200 pages long. Mr. Samsa might correct me on this, though, because in the intervening time I have not kept up with developements that may have occurred.

Incidentally, Skinner conducted experiments on pigeons, as a result of which he claimed that one can induce superstitions in them. Again, he did not bother to investigate brain, mind or whatever. He just studied their reaction to his conditioning. Of course other psychologists disputed that the pigeons were even capable of behaviour from which superstition can be detected, but I could not follow the drift of their argument.

[Mr.Samsa]

Hey Rev, interesting questions. I'll do my best to answer them.

Anyone else can answer this as well, if they know anything about it. I'm thinking pigeons as I recall that Samsa, as a behaviourist, has talked about pigeons and their processes of learning. If pigeons don't fit this example, then substitute rats in or some other appropriate animal.

Looking at the questions, pigeons or any other animal will work fine.

Let's say we've got 50 pigeons and we perform a large enough number of experimental replications.

Pedantic side note here, but if you're interested a lot of these experiments work slightly different to how a lot of people learn how to do scientific experiments. Usually we learn that you need a big sample, then you divide them into two groups (one control and one experimental, or more if necessary) and then you compare differences. However, we often use small-N or within-subject comparisons, which means we can achieve the same accuracy and statistical certainty with only 5-6 subjects. The key is that for the large scale studies we tend to have lot and lots of subject but few data points, whereas for small-N studies we have few subjects but we have lots and lots of data points. The former is good for finding general trends and the latter is good for identifying individual differences - we do the latter because we're trying to find universal laws that apply across every animal, so erasing individual differences with statistical averages defeats the purpose of our research.

Another advantage of this is that not only are the subjects their own controls (i.e. we measure the differences in responses in all the conditions for the best comparison) but because we run the conditions over and over and over again, they serve as their own replications so we get thousands of replications in each experiment. Then we tend to perform multiple experiments to demonstrate a single claim, which is then obviously repeated and replicated by other labs and research.

So, let's start at the end and work backwards, with a broad question about animal learning.
Q1. Are the processes of learning broadly the same between pigeons and humans? As a behaviourist, I'm expecting Samsa to say yes, and as a pretty full on materialist I'm tempted to say that yes, they are broadly the same processes.

Depends on exactly what you mean but yes, the fundamental laws of learning are basically the same across all animals.

So, Q2 is seeking an explanation of the processes of learning shared by the two different species where there appears to be a different level of cognition going on.

To expand: What's going on in the pigeons brain while it is standing there in front of the three buttons? How does past learning manifest itself in the fact that the pigeon after enough trials is going to most likely pick the red button? Substitute for the second scenario: What's going on in the human's brain while we are standing there in front of the three buttons? This is where, IMO, there is a difference in the level of abstraction in the "mind". Human's are conscious. Pigeons aren't? I dunno, but I'm guessing they aren't.

Pigeons are conscious and capable of conscious recall - one good line of evidence for this is the phenomenon of mental time travel. This is the process where organisms recreate past events in their mind and use it to plan for future behavior, so I think the evidence supporting that is relevant to your question.

As for what is going on, basically what you've described is just operant conditioning. The best way to understand how this works is to think of operant conditioning as natural selection within an organism's lifetime (Skinner even explicitly made this link, preferring to call it "selection by consequences"), and so you have sets of behaviors which either get to "reproduce" (i.e. strengthened) or are "culled" (i.e. weakened). So with your three buttons, in that situation two of the buttons are set on a schedule we call "extinction" (again with the explicit link to natural selection) where responses aren't reinforced/rewarded and so the behavior sets directed at those buttons die out. On the other hand, the reinforced behavior is strengthened, so the probability of it occurring again in the future is increased.

Now you're right that there can be differences in the learning of humans and pigeons, and (without recalling any specific data on the subject) I think you're right that humans will learn quicker, and again I think you're right to zero in on past learning playing a role. I'd argue that the reason for this difference is because even those these fundamental learning processes are the same, the capacity for higher order learning can differ depending on differences in things like neural architecture, intelligence, cognitive capabilities, etc. In other words, in all animals when they learn an association it isn't limited to that specific situation or context, they also learn how to generalise. Some animals are better at generalising than others and, for humans, we seem to be better at something called 'concept formation' which is basically just a way of chunking those generalisations together in a more coherent framework. So instead of simply generalising a relationship like "If X then Y", and generalising a second relationship like "If Y then Z", humans will learn meta-relationships between past associations, so we'll simultaneously learn "If X then Z", even though we have no experience with that association whereas most other animals struggle with this (or, at the very least, we have sparse evidence of other animals doing this).

So, back to the human. We possess a fantastic ability to hypothesise, that presumably most other animals don't possess. That is, we can decouple our thoughts/experience of the phenomenal world from creative constructs of a hypothetical reality. For example, I am sitting on my couch looking at the room around me. The TV is there, there's doors and windows and whatever. It's all pretty safe and there's no need for me to be overly worried at this point (unless I was Seth; I'd be fully locked and loaded and have nightvision gear and stun grenades at hand :p ). But I can sit here and look at this same scene and basically invent a tiger into the scene. Thanks to the human ability to decouple our thoughts from reality, I can imagine this tiger growling at me without figuratively and probably literally shitting my pants as would be the case if I couldn't tell the difference from "reality" and my imagination.

Anyway, relating this to the human scenario... We are standing there and we can recall past memories and essentially represent a hypothetical state of the world where we press either the green or blue button and no food comes out. We can then also recall and hypothesise how pressing the red button produces food. And thus we have learnt.

I don't think there's any need to limit hypothesising to humans. The mental time travel phenomenon that I mention above involves some of this, and Hermit provides another good example - superstition. I think this shows that other animals are capable of taking the facts of a situation and inventing possible explanations for it, then acting on those hypotheses. The fact that they can be wrong is particularly interesting and relevant to your question, I think.

Now, that seems all pretty uncontroversial for the human example. But what the hell is going on in the pigeon for it to learn from past experiences? If there is no conscious recall, or at the very least no decoupling and hypothesising, how is it actually learning? There's a couple of possibilities, in my mind. 1. The process between pigeons and humans is exactly the same, and the "conscious" business going on in the human's mind is just a post hoc story we invent to explain how an event happened (i.e. us learning to pick the red button). That still leaves the question of how either of us actually 'learns'. And 2, The process is the same between the species, but the human's consciousness and decoupling abilities allow us to learn quicker than the pigeons. And there's probably a third possibility, and that would be that the process ISN'T the same between the species and there's something totally different going on, probably centred around the differences in conscious awareness.

I think there's another possibility that I don't think is covered by your suggestions there - our learning processes are the same in kind but differ in degree (for the reasons I discuss above). I don't think this requires the "decoupling abilities" you mention or a claim about differences in consciousness, but still predict the results we're trying to explain.

If you wanted to try to answer how operant conditioning works, well that's a bit of a rabbit hole and it's a pretty contentious area at the moment. Originally there wasn't much thought put into it and people just assumed a position called "associationism" - basically one thing occurs around the same time as another thing, so they get linked. Doesn't matter how, they're just linked, end of story. But about 10-20 years ago things got a little complicated when we started finding results that didn't seem to match this view. For example, one basic assumption of associationism is that if an event occurs around the same time as a punisher, then the two should become linked and so the behavior that preceded the punisher will be punished (i.e. decrease). However, there are some experimental designs where different schedules are set up without any signals to indicate what kind of schedule they're responding on, except that they can sometimes press a button that briefly flashes up different coloured lights to let them know if the schedule is a rewarding one or a punishing one.

If we set up the situation so that we can measure their behavior in conditions where the observing response is mostly signalling rewarding outcomes, or mostly rewarding punishing outcomes, or equal amounts of both, we can make different predictions about what we should see if associationism is true. Specifically, if the observing response mostly results in punishing outcomes, then the pigeons should be less likely to choose to observe what schedule they're in, because it's effectively been punished. We don't see that though. Instead we see that the pigeons still choose observing responses even when, in the past, it's been paired with punishers.

New explanations for why this happens are still debatable but some of the research I've done in the past tried to view it in terms of what's called the "information hypothesis". It's actually an old idea that never really took off, but was actually made to be consistent with Pavlov's original research on classical conditioning - it turned out that when he talked about "conditioned stimuli", it was actually a mistranslation and he understood them as "conditional stimuli". It's a subtle difference but his point was that it wasn't based on a form of associationism, and instead it was a signal or cue that allows the animal to make a choice about what they want to happen. So when punishers decrease behavior, it's not because of some simple link between the two things but rather because the animals predicts what is likely to occur in the future and chooses to avoid that. This makes sense when looking at the situation I describe above because there is no reason for them to want to avoid the observing response, as it gives them information about future events and improves their ability to predict and choose what they want. There's a good article here which likens things like reinforcers to signposts.

Anyway, hopefully all that makes sense. I'll clarify or expand on anything that you need me to.

[Mr.Samsa]

Having read a bit about behaviourism almost 40 years ago, my memory is a bit vague, but as far as I can recall it pretty much ignores considerations concerning brain, mind, consciousness et cetera. Instead it focuses on action. Behaviourism is very much process driven, limiting itself to observing actions/reaction sequences. It investigates the 'how' and ignores the 'who' and 'why'.

This isn't quite true. What you're describing is methodological behaviorism, which was the first brand of behaviorism described by Watson. It was considered 'methodological' because it argued that, for entirely practical reasons, we should ignore what's going on in the mind and just focus on what we observe. This was an attempt to try to unify the field and flesh out some solid findings, and aim away from the meaningless speculation that many researchers at the time were engaging in (among a lot of good research too).

The part that most people confuse or misunderstand is that this wasn't an absolute statement. The point of methodological behaviorism was always a stepping stone, and once that groundwork had been laid and we understood enough about behavior (and our technology and scientific methodology advanced enough), we were always supposed to start investigating the mind. That's where radical behaviorism comes in around the late 1920s, which was Skinner's brand and labeled "radical" because it brought the mind back to psychology. The fundamental point of radical behaviorism was that not only can we study the mind and cognitive processes using a behaviorist approach, but that we should. This is the position that the behaviorists follow today and, although some may debate this, it's the dominant methodological approach in psychology for everyone. This is off-topic but I say it for two reasons: 1) even the most ardent cognitivist will agree that basically all methodology in the field is taken wholesale from the behaviorists (we've been criticised for a number of things but never for the quality of our research), and 2) I don't think there is any difference between radical behaviorism and cognitivism - most claimed differences seem to stem from a misunderstanding of what behaviorism is and says.

The process has been given a label: Operant conditioning. You'll find very little discussion concerning what it is that is being conditioned or what enables it to work.

I've talked about this above, but you are half right here. There is an mindset among many researchers (especially in the past) which meant that they didn't delve too deeply into deeper questions like what's behind operant conditioning but I think it's dying out a bit now, and some of the research I discuss above should help highlight that. There's also a lot of behaviorists working in fields like neuroscience, digging at the problem from a different area to expand our understanding of it.

You may find B. F. Skinner's book, published in 1971 and titled Beyond Freedom and Dignity a useful introduction. It won't take long because It is easy to read and not much over 200 pages long. Mr. Samsa might correct me on this, though, because in the intervening time I have not kept up with developements that may have occurred.

'Beyond Freedom and Dignity' is a good book to read but might not help him with his specific questions here, but it's more a philosophical defence of his behaviorist approach rather than a scientific investigation into how behavior works. The gist of the book is just the idea that the folk ideas and language we use to describe psychological processes muddies the water, and to move forward we need a whole new approach to understanding these topics that aren't bogged down with superstitious beliefs or unnecessary baggage.

A better book from Skinner for Rev's questions would likely be "Science and Human Behavior" (there's a free copy available if you just google the title), or for a more modern take on the topic, William Baum's "Understanding Behaviorism" is good.

Incidentally, Skinner conducted experiments on pigeons, as a result of which he claimed that one can induce superstitions in them. Again, he did not bother to investigate brain, mind or whatever. He just studied their reaction to his conditioning. Of course other psychologists disputed that the pigeons were even capable of behaviour from which superstition can be detected, but I could not follow the drift of their argument.

Again, I'd just point out that Skinner was absolutely interested in what was going on in the mind (and he didn't investigate the brain because he wasn't a neuroscientist but was very interested in the research), but the main arguments against his superstition research was that they questioned whether his descriptions of the behavior were accurate and actually a response to "adventitious reinforcement", or whether they just happened to be flapping their wings or whatever and was unrelated to the schedule (the argument is a little more complex than that but that's the basics). There might be some truth to the argument but superstitious behavior has been demonstrated using the same procedure so many times now that it's not really debatable. It's such a standard part of our understanding that we have to include components to our schedules to ensure superstitious respond doesn't affect our results - for example, in an experiment we'll often expose a pigeon to multiple schedules that change every now and then, but since pigeons can be responding for a while without reinforcement for a while on one button, we don't want them to peck the other button and get a reinforcer simply because one was set up there for 1 response, and conclude that they need to tap the first button for minutes and then switch to the second button to get the reinforcer. So we have a thing called "changeover delays" where no matter what the schedule has set up, there will always be a delay when switching buttons so that they don't develop these superstitions.

[pErvinalia]

Thanks for that detailed answer, Samsa. I'm going to take a couple of days to think about it and see if there's any sensible questions I can come up with.

[Mr.Samsa]

Thanks for that detailed answer, Samsa. I'm going to take a couple of days to think about it and see if there's any sensible questions I can come up with.

No problem, take your time.

[JimC]

Mr Samsa wrote:

...I think this shows that other animals are capable of taking the facts of a situation and inventing possible explanations for it, then acting on those hypotheses...

Not disputing this could happen at a basic level, but what they are probably not capable of is turning such a situation into a internal narrative via a language complex enough to be a template, where alternative versions could be compared in advance of future decisions, and the results of such reflection being communicated to others.

[pErvinalia]

Having read a bit about behaviourism almost 40 years ago, my memory is a bit vague, but as far as I can recall it pretty much ignores considerations concerning brain, mind, consciousness et cetera. Instead it focuses on action. Behaviourism is very much process driven, limiting itself to observing actions/reaction sequences. It investigates the 'how' and ignores the 'who' and 'why'.

This isn't quite true. What you're describing is methodological behaviorism, which was the first brand of behaviorism described by Watson. It was considered 'methodological' because it argued that, for entirely practical reasons, we should ignore what's going on in the mind and just focus on what we observe. This was an attempt to try to unify the field and flesh out some solid findings, and aim away from the meaningless speculation that many researchers at the time were engaging in (among a lot of good research too).

The part that most people confuse or misunderstand is that this wasn't an absolute statement. The point of methodological behaviorism was always a stepping stone, and once that groundwork had been laid and we understood enough about behavior (and our technology and scientific methodology advanced enough), we were always supposed to start investigating the mind. That's where radical behaviorism comes in around the late 1920s, which was Skinner's brand and labeled "radical" because it brought the mind back to psychology. The fundamental point of radical behaviorism was that not only can we study the mind and cognitive processes using a behaviorist approach, but that we should.

What I want to know is why so many make this mistake? Why is it that people don't understand that the "radical" in Skinner's behaviourism was related to what came before, not in relation to cognition? So sick of pointing this out to people on the internet (most commonly, Marxists, for some reason).

[pErvinalia]

Ok Samsa, a general bit of discussion before I get around to reading your links.

Let's assume that (via those links and other studies) that pigeons are conscious to a degree. Although, to qualify that, "consciousness" seems to be variously defined, depending exactly on what one is talking about. As a general principle, it seems uncontroversial to me that consciousness isn't going to be an "on/off" phenomenon. It's likely to be a sliding scale of "level" of consciousness. The question is: Is the gap between humans and all other animals (or other groups, such as apes and all other animals, or apes + cetaceans and all other animals) such that it turns out to be effectively an on/off situation? Let's assume that it isn't, given that you've suggested there is evidence that pigeons are conscious, and I would think that "tool use" in some corvids and other birds would also likely suggest consciousness of some degree. An interesting question here would be to define "consciousness" and see where all these animals fit in. To me it would seem that some form of internal representation (a la the "Cartesian theatre"), and a clear sense of identity - that is a self separate from and interacting with the external world - would be necessary. Dunno what behaviourists consider "consciousness" to be. Perhaps you can expand on that, Samsa?

Anyway, if we operate under the assumption that pigeons are conscious, let's take it further back in the evolutionary line and find an animal (or insect) where we would consider consciousness to be absent. Do those animals still learn? If so, what is the process that's going on there, given it seems that both pigeons and humans learn consciously (at least initially before it becomes conditioned and becomes subconscious)? How are these unconscious animals learning?

Does this get back to your point about the "natural selection" of good and bad behaviours within an organism's lifetime? In regards to that, what's the mechanism by which this works? It does sound a bit like a "just so" story. Unless we are talking about ideas/memes, which it does appear are replicable units that can be selected for or against. But in any case, that would seem to imply a high level of consciousness where ideation can even exist.

[Mr.Samsa]

Mr Samsa wrote:

...I think this shows that other animals are capable of taking the facts of a situation and inventing possible explanations for it, then acting on those hypotheses...

Not disputing this could happen at a basic level, but what they are probably not capable of is turning such a situation into a internal narrative via a language complex enough to be a template, where alternative versions could be compared in advance of future decisions, and the results of such reflection being communicated to others.

Sure, to a degree. I don't think they'd be using a form of language to collate and understand internal concepts, which limits their capabilities in comprehending the world and communicating it to others, but I think for my point above I'm just saying that they are running alternative versions of events in their heads and weighing up future decisions, and some animals can communicate these conclusions to others (but obviously not to the same degree as humans). That's essentially just all that social learning is, like with New Caledonian crows where they learn the best way to fish grubs from trees and pass that information on to their offspring, so they don't have to go through the trial and error process.

What I want to know is why so many make this mistake? Why is it that people don't understand that the "radical" in Skinner's behaviourism was related to what came before, not in relation to cognition? So sick of pointing this out to people on the internet (most commonly, Marxists, for some reason).

I have a number of possible explanations but I'm not sure if anyone has looked into it in any rigorous way. One aspect of the problem, in my opinion, is that methodological behaviorism caused quite a stir (especially with Watson's reputation) and so the label "behaviorism" was forever tarnished, practically priming people to reject anything that includes the term. This is compounded by the fact that Watson's behaviorism was also heavily misrepresented. He argued in favour of trying to balance the research into causes of behavior, as up until his day everyone focused heavily on biology/genetics, and so people viewed him as a blank slatist - even though his famous "12 infants" quote goes on to reject that view of him, and the last couple of chapters of his book were dedicated to the importance of instincts (which makes sense, as he was an ethologist who spent most of his life studying innate behaviors).

Those problems could have been avoided by Skinner altogether if he had only chosen a different name for his philosophy, but for some reason he chose 'behaviorism' despite being quite a serious split from the behaviorist philosophy of the time. From what I've seen, most theorists argue that he did this as a head nod or appreciation of the historical progression of ideas but unfortunately it's been viewed as him directly adopting the views of his predecessors. Skinner also didn't make it easy for himself given that a lot of his arguments were very detailed, and required understanding his work from other books or papers for them to make sense, and so isolated comments or statements took on their opposite meanings. For example, his behaviorism is often described as ignoring the mind because he wrote regularly about how behaviorism is the idea that behaviors can be explained without recourse to the mind - however, that statement comes from an argument he has about levels of explanation, which has nothing to do with whether the mind or cognition contributes to behaviors. In other words, he was simply saying that behavior can be studied at the behavioral level and for it to be scientific, in the same way that chemistry can be studied at the chemical level and for it to be scientific. That doesn't mean cognition or neuroscience doesn't play a role in behavior, or that physics doesn't play a role in chemistry, just that you don't need to study those topics for research above that level to be scientific.

Ok Samsa, a general bit of discussion before I get around to reading your links.

Let's assume that (via those links and other studies) that pigeons are conscious to a degree. Although, to qualify that, "consciousness" seems to be variously defined, depending exactly on what one is talking about. As a general principle, it seems uncontroversial to me that consciousness isn't going to be an "on/off" phenomenon. It's likely to be a sliding scale of "level" of consciousness.

I can happily accept that for now, but I might suggest that what you're getting at might be something closer to the distinction between sentience and sapience?

The question is: Is the gap between humans and all other animals (or other groups, such as apes and all other animals, or apes + cetaceans and all other animals) such that it turns out to be effectively an on/off situation?

You're right that I'd debate this. So I'm happy to accept that consciousness is on a sliding scale, but I don't think the scale slides so much as to be overly relevant to the questions you're asking.

Let's assume that it isn't, given that you've suggested there is evidence that pigeons are conscious, and I would think that "tool use" in some corvids and other birds would also likely suggest consciousness of some degree.

Certainly, but I'd go further and suggest that almost any animal capable of operant conditioning is conscious to some degree.

An interesting question here would be to define "consciousness" and see where all these animals fit in. To me it would seem that some form of internal representation (a la the "Cartesian theatre"), and a clear sense of identity - that is a self separate from and interacting with the external world - would be necessary. Dunno what behaviourists consider "consciousness" to be. Perhaps you can expand on that, Samsa?

That would be quite a high level concept of consciousness. Usually it just refers to a state of awareness or ability to experience, and I'm not really aware of any behaviorist approach that defines it as anything different. However, what most behaviorists will do is argue that broad labels like "consciousness" are problematic for the reasons you've sort of been scratching at - they mean different things to different people, and instead they'd argue that we should clearly define what aspect of consciousness we're interested in and study that. So if you're interested in something like internal representation, they'd ask "How can we study this and find clear evidence for or against it?", and then get to it. For example, with that example, they might argue that theory of mind requires some degree of internal representation and then go on to show animals completing theory of mind tasks.

They'll accept "consciousness" as a useful umbrella term, in the way Gilbert Ryle describes for "intelligence" in "The Concept of Mind", but when trying to make specific claims about it they'd ask for a more specific testable operationalised term.

Anyway, if we operate under the assumption that pigeons are conscious, let's take it further back in the evolutionary line and find an animal (or insect) where we would consider consciousness to be absent. Do those animals still learn? If so, what is the process that's going on there, given it seems that both pigeons and humans learn consciously (at least initially before it becomes conditioned and becomes subconscious)? How are these unconscious animals learning?

Like I touch on above, I personally don't think it's possible to learn by operant conditioning without consciousness. To me it requires not only the ability to experience something, but to be able to also 'replay' that experience in the future to make a decision at a later time. Classical conditioning is debatable, and on certain days I'd argue even that requires some ability to experience and so I'd say if there's no learning then there's no consciousness. But on other days, I'd say that the earliest common ancestor without consciousness would likely be one that could only learn through classical conditioning.

Does this get back to your point about the "natural selection" of good and bad behaviours within an organism's lifetime? In regards to that, what's the mechanism by which this works? It does sound a bit like a "just so" story.

I'm not sure how it's a just-so story, I'm not saying that the behaviors are evolved (necessarily, they might be but that's not my point). The comparison to natural selection is just in the process by which they work - i.e. selection by consequences. So behaviors that aren't culled off survive, and spread and build. That's just a description of what happens, so it can't be a just-so story.

Unless we are talking about ideas/memes, which it does appear are replicable units that can be selected for or against. But in any case, that would seem to imply a high level of consciousness where ideation can even exist.

The idea of "memes" is sort of similar to what I'm talking about (in a very general sense) but it's not limited to "ideas". If I put a rat in a box with a button that sometimes gives it food, then multiple behaviors will survive and die across an experimental session. For example, cleaning behaviors will die off in favour of pressing the button. It might be a little confusing to compare this concept with memes though - memes are treated as pseudoscience, and even the journal dedicated to documenting all the research on it had to make a statement explaining that they were closing down because there was no evidence to support the concept.

[pErvinalia]

It sounds like a potential "just so" story, as there doesn't seem to be a conceivable way to know what goes on internally regarding "selection" of behaviours. How do you actually explain that? Why should ideas/behaviours be "selectable"? By what mechanism is it selectable? You could just say as well say that the inner homunculus made a decision about which ideas/behaviours get chosen.

[Mr.Samsa]

It sounds like a potential "just so" story, as there doesn't seem to be a conceivable way to know what goes on internally regarding "selection" of behaviours. How do you actually explain that?

Why would we need to know what goes on internally? We're talking about measurable behaviors, we observe what behaviors survive and which don't. Internal behaviors follow the same rules as external behaviors so we can investigate and measure these things in that respect as well, but we don't need to talk about "internal" aspects for the issue being discussed.

Why should ideas/behaviours be "selectable"? By what mechanism is it selectable? You could just say as well say that the inner homunculus made a decision about which ideas/behaviours get chosen.

Behaviors should be 'selectable' because one behavior will necessarily win out over other behaviors, given that there are inherent limitations on what behaviors can be performed at certain times - that is, you can't perform one behavior and simultaneously perform an incompatible behavior. This is why all behavior is viewed as 'choice behavior', as whenever you are behaving you are in effect choosing not to behave otherwise.

You could say that a homunculus makes the decision about what gets chosen, but that would be adding an explanation which needs to stand or fall on the evidence we can find to support it, whereas the selection of behaviors is just a description of what happens, not an explanation or a theory that needs supporting.

[pErvinalia]

It's not really the choice that is the problem, it's the use of "selection" in a self reinforcing context (that is "evolutionary", but in the context of individuals and within single lifetimes). You have no way of knowing that there is even a selection mechanism of this sort. That's why it seems like a just-so story. There's all sorts of mechanisms one could invent to explain how behaviours persist.

[mistermack]

I'm not really joining in this discussion, I'm just sharing something that I've just seen.

Where I live backs onto a school. Gloucester has a big population of Seagulls.
Every day, at 11 am and 1pm, the gulls arrive and start circling overhead. Crows too.
Then, when the kids come out with their packed lunches, the birds fight over the scraps that the kids throw away. (messy little gits).

Today is Sunday, and it's 1pm. They are there, as usual, circling overhead, looking for the kids to come out.
They obviously have a daytime clock, to tell them when it's time that food will be available, but can't cope with our five-days-on two-days-off week.

I don't think that it's likely that they THINK about the availability of the food at certain times. They just have a feeling that if they go to this place, they will be able to feed. I think that they probably get that feeling at certain times, without having to work it out mentally.

[pErvinalia]

Too late to edit...

Just to add to that, in evolution the selection mechanism is death from environmental interaction. What's the selection mechanism in the mind for behaviours? It doesn't seem too different to memetics, which you discount.