What you see with your own eyes vs what scientists claim

And then later, If a walnut has 100 energy, and the probability of losing prey looks like this:

metres              probability  of losing
5                         90
4                         45
3                         15
2                         8
1                         6

And if the number of drops required looks like this:

metres              number of drops required
5                         1
4                         2
3                         5
2                         7
1                         10

Then the probability of still possessing the walnut after it cracks open:

metres                probability
5                         10
4                    0.552=30
3                    0.855=44
2                    0.927=56
1                    0.9410=54

So, on average, a walnut dropped at 5 metres nets 10 units of energy (10% of 100). Likewise, dropping the walnut from 1 metre will average out as 54 units for each walnut. The optimum height to drop is 2 metres, since each walnut then effectively gives 56 units of energy. If 2 metres is chosen then the subject has maximised the amount of energy they get from the walnuts.Obviously this is a simplified account, but that is how it ties into the idea that crows are maximising energy.Edited by Modulous, : No reason given.

Right, so we're back to

quote:

---

If a bird scientist observed humans and thought it worth mentioning, with no evidence, that one possible reason that certain humans can play good poker is because they are on drugs or brain damaged - they would get an appropriately amused twitter I'm sure you'd agree.

---

Last time you disagreed because of performance enhancing drugs in athletics. I think we can now agree that a drug that makes you good at poker and a drug that builds muscle mass to increase speed/strength are different categories. They tested prey breakability. Their model predicted that the optimum strategy for harder nuts is to drop them from higher heights. The crows dropped harder species of nut from higher heights.They tested substrate hardness. They predicted that the optimum strategy for dealing with softer surfaces is to increase the height to drop from. It also predicted that the optimum strategy is to prefer harder surfaces. The crows dropped from higher onto soft surfaces and preferred hard surfaces.They tested the height of successive drops. They predicted that the height from which the walnut is dropped on successive drops will lower as the probability that the walnut will break goes up. They observed this behaviour in the crows.

quote:

---

At Birch Lane 11.9% of 337 English walnuts dropped once were dropped again by the same crow (mean = 3.42 1.87 drops). The maximum number of repeated drops observed at Birch Lane was 17, although at other sites crows dropped English walnuts up to 50 times before cracking them (data not shown). As predicted by our model, drop height decreased significantly with successive drops (Figure 3). Attempted kleptoparasitism did not change with successive drops (df = 4, Kruskal-Wallis H = 4.0, p =.41), indicating that this variable is unlikely to explain the relationship between drop height and successive drop number.

---

They tested prey mass and breakability. They predicted that the optimum strategy is to drop heavier walnuts from lower heights. The crows did not follow this behaviour.They tested prey loss. They predicted that as the risk of kleptoparasitism increases the optimum strategy is to lower the height to drop. This behaviour was replicated by the crows.Thus, with the exception of taking mass into consideration, the crows followed the optimum behaviour in adjusting height to account for various variables, maximising the chances of getting at the sweet nut, and thus maximising the energy per walnut. The paper did not discuss the amount of energy the crow expends, though that would be interesting too. They were talking about the amount of energy gained from the walnuts. I was comparing my quotes with theirs to show how I didn't seen any contradiction with what I said with what the paper showed. Sorry for any confusion.Edited by Modulous, : No reason given.

No, they are for illustrative purposes. A simplified account as it were. Simply to show what it means to maximize the energy obtained from each dropped food item. I'm not married to the numbers, if you have noticed a problem with them, by all means point it out. Hopefully any mistakes present don't get in the way of the actual illustration.

Yes indeed, but they can't provide you with the optimum way of playing poker. They can help you play poker better by increasing your mental faculties, but they can't help convey any knowledge of optimum strategy - they only assist in carrying out the learned optimum strategy. Energy gain (and flight cost) were part of the model. It is in the
paper that discusses the model.

quote:

---

In the simplest situation, an individual has a food item that it cannot lose, and the probability of breaking open the food item is independent of the number of times it has been dropped previously. As a necessary simplification for the model, we assumed that the individual has a set number of drops, D, in which to break the food item open. If the food item is not open after D, the maximum number of drops, the bird gets no reward from the food. Hence, the expected energy value from the food item at D is E(D) = 0

---

It goes on to discuss more complex situations.

It might be tricky to do the 'translation' because game theory of chess is much trickier. However, I'd be very surprised if brain damage or drugs conveyed the best strategy on someone. They could aid in the mental search process to look for optimum strategy, but one would need to have some idea of what optimum strategy might result in (favourable position, material advantage etc), and in chess veterans this is usally done by piecing together general rules of thumb tactics together with slight modifications combined with general strategic rules of thumb...so the reason humans are good at chess is mostly because of learned behaviour rather than drugs or brain damage. Kind of. Instead of doing this, they calculate a probability of cracking at any given dropped height.

quote:

---

the food item may break with probability P_b (h), with the individual receiving the net energy E_break. Alternatively, the food item may not break [with probability 1 ” P_b(h)], and the individual receives the net energy of E_intact plus the maximum energy it can expect from having the food item from drop d + 1 to D...

​

The results consist of a single optimal height, h*, that satisfies Equation 6 for each d.

---

They can then examine the model and see how certain variables affect this optimal height. See Table 2 for an simple summary. Edited by Modulous, : No reason given.

It requires more than that. 'envisage' is just another way of saying learn. Yes, drugs can boost learning performance, they don't convey unlearned optimum solutions upon their taker. True enough. Then again, freezing is hardly the level of behaviour we were talking about with crows. They engaged in a specific set of behaviours that were all the correct way to approach an opimum height. If you took a drug that meant you were able to play poker or chess without first having learned how to play poker or chess - then we'd get somewhere. Maybe it is possible, but once again, when we look at a group of humans playing chess we don't conclude that the evidence obtained by watching them play with fairly good strategy suggests that chess is either a learned, evolved or drug-induced behaviour.If we gave our freezing drug to a collection of humans and watched them deal with a number of frightening situations we can try and determine what the optimum behaviour in each situation is. Run away or freeze. We can compare the drugged humans to some normal humans and then decide which is the optimum behaviour - and if we find that the optimum behaviour is to freeze then I agree that some interesting questions should be asked. No, that is not the question. And why are you excluding brain damage? Brain damage could theoretically specialise a human brain into chess playing mode. You could also include cybernetics. The answer to your question is obviously yes. If a drug could improve speed of cognition then it could theoretically improve your game.However, the question in issue is more complicated. I'll revert to the easier poker, feel free to do any translation you think is needed.In poker we can do fairly easy calculations to approximate the optimum action. This action is usually a combination, for example: Raise 70% of the time, fold 25% of the time, call 5% of the time.Once we have used our model to determine the optimum we then observe some animals playing poker. We find that under the circumstances above, the animals raise 70% of the time, fold 25% of the time and call 5% of the time.How did they know how to do that?One possible way is that they have learned how to play poker, and have constructed a mental model that matches our own mathematical model. They've learned what the optimum strategy is and now they follow it.Another possible strategy is that the model is part of their mental makeup: hardwired into their brains. This kind of structure in the brain is almost certainly shaped by natural selection (and we'd have to explain how playing poker might prove advantageous...).The third possible you raised is that the model comes by ingesting the drug, that the drug seeks to rewire the brain or redirect the signals in such a way as to mimic the optimum model.There are a few other possibilities, including alien technology, ghosts possessing the animals and so on and so forth. As I pointed out previously, if 'we' are birds and we raised those possibilities in a science paper and there is no body of work or evidence to back it up, we'd get amused twitters all around. Would you be terribly surprised if I was getting the same impression from you? I had expected that you had reviewed the model, so I didn't realize you wanted me to explain the whole model to you. Perhaps it would be wise for you to review the model to your satisfaction before commenting about it. I await your respsonse.

Yes, my illustration also ignored mass of the prey, number of times previously dropped, substrate hardness, acceleration due to gravity and probably more things to boot. That's why I didn't call it a 'complete model' but instead I said it was simplified. If you weren't sure what simplified means, it means that the model ignores various factors.

Drugs don't actually produce the solutions though. The solutions are the results of a reasoning process. Reasoning how to do something by constructing an internal model is a part of learning. Increasing the capacity to learn doesn't change that it is a learned behaviour. Unless you propose a drug which can tell you how to play good chess with no experience of how to play chess. When you get better at chess it is because you are learning to play better chess.Reading a book would also help you envisage new solutions of which you have no experience. That is also learning. If the decision to freeze isn't just an incidental effect, but the result of an analysis that would not be possible without the drug then you'd be right. That's the kind of thing I'm looking for. If instead it was the more likely case that the 'what do I do' signal was answered with 'I don't know' and the 'I don't know' reaction was redirected to 'do nothing' rather than 'run away' then this wouldn't be the kind of thing I'm looking for. Well, if it simply modified the behaviour in a small way which happened to be optimum in that one case - then that wouldn't really be what I was looking for. If it managed to modify behaviour to make a large number of poker related decisions move towards the optimum - that would be something interesting. Well the paper we looking at initially was examining a "If loss varied with height" scenario. What you describe is for "loss does not vary with height" scenario. Different scenarios with different outcomes. I seem to remember that you recognised that "The number of drops required to break the nut was missing from their test as an independent variable. But it is a critical variable in measuring energy expended.". Do you still think that their test is flawed having viewed the model?

You've done well Percy, I think I started losing the point back in Message 31. The only way I can get it to work is think that his eyes can lead him to a superior model of avian-prey behaviour than these scientists. Maybe he's just trying to show that scientific papers are not perfect documents of the living word of Science, and that that lends strength to his position but that only works if we put a huge amount of focus on a small possible flaw in some science paper and ignore the fact that the scientists are also human which demonstrates conclusively that humans can be flawed in their observations/conclusions about observations.I'm hoping I'm wrong, and that we might get to interesting topics about the possible problems of the conservatism in science which don't plague the more credulous 'guy on the street' from this discussion.

Reasoning can be seen for our purposes as drawing upon experience, and trying to create a coherent narrative or model. If reasoning does not draw upon experience (pure rationalism that is), then it is very unlikely to happen upon a way to an optimum chess strategy without any experience of chess or games or other strategies etc.Gaining new knowledge due to past experiences is learning. Acting on that new knowledge would surely then be classed as learned behaviour.I did not come to the conclusion that the solutions are the result of reasoning - I discussed both options in my post. Their model predicted that at no loss the optimum height was at 16. If loss probability varied with height it was lower than this height. Are you thinking that loss probability might possibly go down as height increases? That scenario is (as you have already observed) not part of their model. We discussed why that is reasonable already, and indeed is observed behaviour in the birds. Yes, they used the evidence of their eyes which showed them that the probability of losing something increases the longer you leave it unprotected. Have you a superior probability function other than the one used? If we use it, do you think it will successfully predict the behaviour of crows?

Indeed, or we can theoretically do it by surgery or science fiction like with cybernetics or matrix-like downloads. Yes, I would say it was essentially impossible to form purely rational chess strategies once you have experienced chess. Any strategy will be even tainted by the fact that you know it takes place on a 64 square board. Of course, you could forget everything about chess and go back to attempting to discover a good chess strategy without any remembered experience of it. Yes. I see it. They show what happens in the fixed loss scenario and what happens in a increasing risk of loss with height scenario. The scenario that they don't cover is a decreasing risk of loss with height. That was the scenario I was referring to: "Are you thinking that loss probability might possibly go down as height increases?", and as I rightly observed: That is not in the model. Let me go back the original argument: In the simple model I presented to you earlier, a similar thing happened. I designed it so that if you were at 1m you would need to increase your height to maximise your energy. Different reasons, but the idea that the height can be increased to maximise energy is not news to me.Now, the graph shows that the optimum drop height can rise with increases to kleptoparasitism probability if the risk of loss does not increase with height.So I thought that maybe you disagree with them picking a risk that linearly increases with height, that might be an interesting avenue of thought and that's why I suggested you provide your modified equation for P_L(h).Edited by Modulous, : No reason given.

Well, that's how philosophy often ends up. I was contrasting pure rationalism with rational empiricism. I think that rational empiricism is perfectly possible in the case I described, but pure rationalism with no empiricism isn't. Still, you're right in that it is a philosophical issue that isn't going to be quickly resolved, so we might as well just leave it there. "It" is not in their model, where 'it' is P_L going down as height increases. That simply isn't there at all. Their model covers three possibilities (wrt kleptoparasitism): P_L going up linearly as height increases, no P_L at all and P_L not increasing with height. It doesn't cover P_L going down as height increases - it really really doesn't.I'm sure you don't think it does, and you are just misunderstanding what is being talked about. I'm just trying to clarify now what I actually said. Yes, I anticipated that might be what you were trying to say, and that is why I wondered if you had a preferred way of calculating P_L(h) you would care to share with us. More importantly to the point, does your method of calculating P_L(h) do better at predicting the relative behaviour of the crow-walnut system than theirs does in Paper II? Part of their model is the linear increase in P_L with increased height, so you will need to modify their model to get results different than is shown in that graph.So far, you seem to be saying thatThe paper does not just suggest that the crows learned the behaviour or that it is an evolved unlearned behaviour, but you also suggest that it is possible that it is an unevolved unlearned behaviour. At this time you have been unable to come up with an unevolved unlearned behaviour that is actually suggested by the contents of the paper.The paper makes an anthropocentric gaffe in not considering that the crows engaged in pure rationalism to move towards an optimum height. This is a philosophical issue that, whilst unresolved is mostly not held by the philosophical community. Ethologists are not going to start postulating it as a suggested possibility any time soon.The paper has some kind of potential error in one of its probability calculations. No better calculation has been presented.Edited by Modulous, : No reason given.