by twistor59 wrote:Thanks for the response. Yes, the question I was interested in was that of fixed action patterns, though I wasn't familiar with that term.
So, staying with the gull example, basically the genome contains the instructions that will build a bird brain which will contain an initial neural net which develops into something which produces the behaviour "peck at a red dot" ?
I guess the difficulty I have is understanding the information content of these behaviours. What I mean is that there is information in the behaviour in the sense that it's
"peck repeatedly at a red dot", and not
"peck repeatedly at a blue dot"
"peck repeatedly at a green diamond"
"turn round once and tweet when you see a black dot"
etc....
Presumably each one of those alternatives could have been a potential neural net, so the genome could have coded for them. There are huge numbers of those alternatives, so choosing
"peck repeatedly at a red dot"
must use up a huge proportion of the tiny (only 1-2 billion base pairs) genome.
That's the bit that's still counter intuitive for me.
I don't quite understand why you think there would be neural networks set up to stop other behaviors.. If that were the case, then I'd agree that such a genome would be infinitely massive, but as far as I'm aware it's unnecessary. The kelp gulls only have a neural network which is "triggered" by a red dot, so this means that whenever they see a red dot they 'instinctively' peck at it - however, this doesn't mean that they can't peck at other things. In fact, although I haven't read any studies on it, I imagine if you presented them with different coloured dots you would still get some pecking behavior as a result of stimulus generalisation, which basically occurs as a result of the sample stimuli being similar in some way to the exemplar stimulus. You find this when you train an animal to, say, peck at a red button and then you test it with other colours; there is some responding on other colours but this level of responding decreases as you move further away from the example stimulus. So orange and yellow buttons might produce a fair bit of responding, but less than red, whilst blue and green buttons would produce little to no responding as they are the furthest away from red.
In simple terms, the "peck at the red dot" behavior is an exception to how behaviors normally come about. We don't have neural networks designed to determine every behavior we should and should not engage in.
my_wan wrote:Mr.Samsa wrote:3) be elicited by a specific stimulus in the environment
I agree with your issues with 'instinct' completely, and fixed-action patterns sums up what I generally refer to as instinct quiet well. My perspective comes from seeing it on an unbroken spectrum, where instinct is merely a useful definition, like speciation. At some level even the distinction between instinct and an inorganic chemical reaction sequence breaks down, except for how we choose the definition to apply. As I'll explain it goes the other way to. My interest in evolution, biology, behavior, etc., is geared more toward applications, which also includes preservation.
I just think it has taken on so many different (and even conflicting or contradictory) definitions over the years that it seems to be a pretty troublesome term to try to hold on to. I agree that there isn't a fine line between innate and learnt behaviors, but I don't think the term "instinct" helps with that.
my_wan wrote:Take nest building and the definition 3) above for instance. None of the definitions given provided for a variable dependence on the hormonal state of an organism of a given species. In fact it was specifically defined as environmental stimulus. So what of fixed-action patterns that are only expressed when certain hormones are present, like those that vary with an estrous cycle? I generally take this as instinct myself, but it opens a Pandora's box once you allow such contingent-action patterns as a stand in for the definition of instinct. We could say estrous is merely a mechanistic certainty of the species, but estrous often has environmental triggers like instinct. In some sense you could call learning a biological response to the environment, like estrous sometimes is, and the learned behavior itself is a change of contingent-action patterns, like nest building is to environmentally induced estrous. I don't object to the usefulness of the definition, but in the big picture I just see such definitions as lines drawn in the sand. The only real difference is just the level of subtlety of the feedback between the environment and the organism, for which there is no real dividing line.
Good point, and you're actually right, in a lot of research the definition of "environment" includes the biological workings of an organism so the estrous cycle would be considered an environmental stimulus. And as you say, this blurs the line but it isn't too big of a problem as these dividing lines are mostly there for convenience in behavioral research. As I mentioned earlier, there is no physical difference between a reflex and a fixed-action pattern, but it's incredibly useful to frame it as if there is.
John P. M. wrote:
However - as nice as this may sound (if indeed it does - I may be talking gibberish, but to me it sounds reasonable right now), there are behaviors that are way more complex, and in which the animal must perform many many tasks in succession for the entire behavior to make any sense and to have an advantage, like building an intricate nest.
As you can see though, there is a factor of hit and miss involved, and it seems they both have the innate behavior, but also have to "practice", or rather try, try again when they miss.
Indeed, this is an important point
Just because something is innate doesn't mean that it cannot be improved with practice, and nor does it mean that the "successful" form of the behavior that we see does not require some fine tuning. The best example I can think of for this is the tool making behavior of the New Caledonian crow. Through extensive research that involves raising chicks from birth without any interaction with their parents, they found that the crows have an innate ability to build grub fishing tools from pandanus leaves. Here's a diagram that shows the basic creation of the tool (sorry the image is so small):
(From here)
They cut out a part of the leaf in an incremental pattern because the holes they 'fish' in become increasingly smaller, and they use the barbs on the outer side of the leaf as tiny hooks to catch the grubs on.
Anyway, what they found was that even though the tool making behavior was innate, it was also flawed without cultural learning. What happened was that the orphan crows could create the basic shape of the tool and get most of the aspects right, but they cut it out of the wrong side of the leaf - as such, they didn't have the barbs and their tool was much less successful than the usual form of the tool.
With further testing, they found that the crows had formed numerous different cultures across isolated of New Caledonia where each population had created it's own "brand" of tool, some specifically to tackle different problems, but some just as a result of a different creation process which had been passed on throughout generations.
The point is, this is an example of an innate behavior which requires a strong learning component - and I think the same is probably true of nest building. (I think even song learning in at least some birds depends in part on learning the song of their parents).
