And by far the greater part of animal communication—outside of laboratories and apart from human tutelage—is instinctive rather than learned.

Of course, but certain instincts are crucial to our ability to learn language. Chomsky proposed that notion in the 1960s, and I think Pinker's work has confirmed it beyond reasonable doubt. Our communicative abilities differ from those of other animals by a degree that constitutes a difference in kind, but we could not begin to communicate the way we do without those instincts.

let us adopt as the minimum meaning that can be agreed to by all parties, the formula proposed by Donald Hebb: namely, that a pattern of behavior can be called innate or instinctive insofar and only insofar as it is "species-predictable," which is to say, in Lorenz' words, "ubiquitous in its distribution" among all members of the species without exception.

I think that will work for me for the time being. If it causes a communicative problem between us, we can revisit the issue at that time.

Has this observation been falsified by any later research?

I can't figure out which specific observation you're referring to.

Note that the distinction between innate, unflexible language is very different from the kind of freedom we humans have with language.

That there is such a distinction is entirely consistent with everything I believe on the subject.

What Adler is constantly concerned about is whether the difference between (i) human use of language and (ii) any other creaturely use of anything that could possibly be called language, is quantitative, or qualitative. If qualitative, then one must account for the emergence of it. One cannot just say 'emergence'; the word is not a magical incantation.

A proper accounting would require a level of scientific expertise in several disciplines that I have not attained. I don't think that means I'm unjustified in believing that natural selection suffices to account for the observable differences between our linguistic capabilities and those of other species.

I can get as formal as you would like

Formal or informal, what I both like and need first is intelligibility. Some of what you've been writing is hard for me understand.

I've yet to read Wittgenstein, although I've read quite a bit about him (primarily via David Braine, Charles Taylor, and Richard Bernstein). My guess is that the Tractatus came from the concrete, analytical, and precise part of his brain, while the Philosophical Investigations came from the speculative, analogical, fuzzy part of his brain. Let either run too far away from the other and you've got problems! I'm very interested to see how good my guess is. :-)

I had to read Wittgenstein's Tractatus while working on a paper about the history of logical positivism, because he was a major influence on that movement. I'm not quite sure yet what to make of his ideas, and I understand that he himself retracted some of them in his later years.

I'm not saying any such creature actually exists. I'm just saying that if there were one, whether it was a product of natural selection or human engineering, we would have no good reason to believe it was not as self-aware as we are.

Ahh, ok. I will have to contemplate that, and compare & contrast it to The Reality of the Unobservable: Observability, Unobservability and Their Impact on the Issue of Scientific Realism (NPDR review), as well as Grossberg 1999 The Link between Brain Learning, Attention, and Consciousness (partial tutorial).

I'll have to get to the rest of your post later.

I look forward to it! As you can perhaps see, I've gotten very interested in language as of late. I'm especially interested to see how Wittgenstein's private language argument interacts with the 'language' that is seen in non-homo sapiens mammals.

But is your italicized bit correct?

In what respect? I'm not saying any such creature actually exists. I'm just saying that if there were one, whether it was a product of natural selection or human engineering, we would have no good reason to believe it was not as self-aware as we are.

I'll have to get to the rest of your post later.

Like Dennett, I reject the possibility of philosophical zombies. I believe that for any creature whose behavior is, in every observable respect, indistinguishable from ours, actual self-awareness, self-consciousness, self-reflection, etc., is necessary.

But is your italicized bit correct? See what Mortimer Adler said in his 1967 The Difference of Man and the Difference It Makes:

    And by far the greater part of animal communication—outside of laboratories and apart from human tutelage—is instinctive rather than learned. Konrad Lorenz stresses this point.

    Animals do not possess a language in the true sense of the word. in the higher vertebrates, as also in the insects, particularly in the socially living species of both great groups, every individual has a certain number of innate movements and sounds for expressing feelings. It has also innate ways of reacting to these signals whenever it sees or hears them in a fellow-member of the species. The highly social species of birds such as the jackdaw or the greylag goose, have a complicated code of such signals which are uttered and understood by every bird without any previous experience. The perfect coordination of social behavior which is brought about by these actions conveys to the human observer the impression that the birds are talking and understanding a language of their own. Of course, this purely innate signal code of an animal species differs fundamentally from human language, every word of which must be learned laboriously by the human child. Moreover, being a genetically fixed character of the species—just as much as any bodily character—this so-called language is, for every individual animal species, ubiquitous in its distribution.[5]

    Lest there be any quibbling about the words "innate" and "instinct," concerning the meaning of which American behavioristic psychologists do not see eye to eye with such European ethologists such as Tinbergen, Thorpe, or Lorenz, let us adopt as the minimum meaning that can be agreed to by all parties, the formula proposed by Donald Hebb: namely, that a pattern of behavior can be called innate or instinctive insofar and only insofar as it is "species-predictable," which is to say, in Lorenz' words, "ubiquitous in its distribution" among all members of the species without exception.[6] (115–116)

[5] King Solomon's Ring, pp. 76-77.
[6] See Donald Hebb, A Textbook of Psychology, pp. 123–126, and 129–130, esp. p. 126. With regard to the differentiation between innate and learned behavior, see N. Tinbergen, The Study of Instinct, Konrad Lorenz, Evolution and Modification of Behavior, Adolf Portmann, op. cit. [Animals as Social Beings], Chapter 5; Irenäus Eibl-Eibesfeldt, "Experimental Criteria for Distinguishing Innate from Culturally Conditioned Behavior," in Cross-Cultural Understanding, ed. by F. C. S. Northrop and H. H. Livingstone, 1964, pp. 297–307.

Has this observation been falsified by any later research? Note that the distinction between innate, unflexible language is very different from the kind of freedom we humans have with language. As I recently said to Jonathan Pearce:

LB: I think it will be important to take into account Wittgenstein's private language argument. What Adler is constantly concerned about is whether the difference between (i) human use of language and (ii) any other creaturely use of anything that could possibly be called language, is quantitative, or qualitative. If qualitative, then one must account for the emergence of it. One cannot just say 'emergence'; the word is not a magical incantation.

I can get as formal as you would like, here; for example, from David Braine's 2014 Language and Human Understanding: The Roots of Creativity in Speech and Thought:

    It is primarily this informality and flexibility in the use of words which enables us in speech, in Von Humboldt's words, to make an "infinite use of finite means"—made possible through the interplay between language-possession and language-use. (8)

One can also consult his 1992 The Human Person: Animal and Spirit, which can be seen as building toward his recent book.

Perhaps one could rephrase the question here: are rhesus monkeys true moral agents, or more like philosophical zombies? Perhaps you reject the dichotomy?

Like Dennett, I reject the possibility of philosophical zombies. I believe that for any creature whose behavior is, in every observable respect, indistinguishable from ours, actual self-awareness, self-consciousness, self-reflection, etc., is necessary. You can have, and in most species with brains you do have, instinct without any capacity for self-reflection. But in those that have it, ourselves included, self-reflection does not replace instinct but only supplements it. There is much that we do that we could not do without self-reflection, but that doesn't mean instinct is uninvolved in our doing them.

There is also the question of whether rhesus monkeys could ever transcend the 'us' vs. 'them' dynamic, which seems very similar to the tribalism we see in the ANE, as well as in parts of the OT (but only parts!). Humans figured out that slavery was wrong. Could rhesus monkeys do the same?

I didn't know they could enslave other rhesus monkeys. If they do, then whether they can get over it depends on the extent to which it is purely instinctual and how much is due to their conscious thought. And whether we have abolished slavery even yet depends on how narrowly you define the word. What we have unambiguously abolished is chattel slavery. There are plenty of other ways to make people work for you when they'd rather work for themselves.

Thanks for doing this. I have a long response drafted, but I would first like some clarification.

1. What is the difference between instinct-driven "willingly choose" and self-reflection-driven "willingly choose"? You key in on that:

As for P(E|~H), here we're asking: How probable is it that this evidence would exist if the hypothesis were false? If monkeys don't experience anything relevantly like human empathy, then how likely is it that, notwithstanding their lack of empathy, they would act as if they did have empathy?

This is very reminiscent of Daniel Dennett's intentional stance. Perhaps one could rephrase the question here: are rhesus monkeys true moral agents, or more like philosophical zombies? Perhaps you reject the dichotomy?

2. There is also the question of whether rhesus monkeys could ever transcend the 'us' vs. 'them' dynamic, which seems very similar to the tribalism we see in the ANE, as well as in parts of the OT (but only parts!). Humans figured out that slavery was wrong. Could rhesus monkeys do the same?

but what are the priors, how are you doing the updates, and what are the posteriors?

I'm not quite clear on what you mean by "updates" or "posteriors," so I'll just go through the analysis starting at the beginning.

First a note about the particular formula I’m using. It is:

P(H|E) = P(H)*P(E|H) / [P(H)*P(E|H) + P(~H)*P(E|~H)].

The formula usually given is simpler than this, but some sophisticated algebra demonstrates that they are equivalent, and this one makes it clearer why the formula works as advertised. And keeping in mind that P(~H) = 1 – P(H), there are only three remaining variables to which values need to be assigned: P(H), P(E|H), and P(E|~H). And one more thing: For every term in this equation, there is an implicit “given background knowledge.” Written in full, every term would have a “.B” appended. Since it appears in every term, we can simplify the writing a little by omitting it, but we don’t get to pretend it isn’t there. Now let’s proceed.

We have a hypothesis: Monkeys experience concern for certain of their conspecifics that is, in some relevant sense, like the concern that we humans experience for certain of our conspecifics. We can say for short that they experience empathy.

We have some evidence: the aforementioned experiment. To apply Bayes' theorem, we need estimates for three variables, all of them based on our background knowledge. The variables are the prior probability of the hypothesis, P(H), the probability of the evidence assuming the hypothesis is true, P(E|H), and the probability of the evidence assuming the hypothesis is false, P(E|~H). And our background knowledge, essentially, is everything, relevant to the hypothesis, that we knew or thought we knew before the evidence was discovered, i.e. before the experiment was conducted.

So we start with P(H). How much credence should we have given the hypothesis before we found out about this experiment? Based on everything I thought I knew about the evolutionary history of Homo sapiens, I would have judged it to be more likely true than false. But let me not stack the deck. Suppose I had no particular reason to believe it. Even so, I would insist that I had no particular reason to disbelieve it, either. And for any proposition, if I have neither more nor less reason to believe it than disbelieve it, I must consider it to have a probability of 0.5, and so that is my P(H).

Now for P(E|H). If the hypothesis is true, then we should have a high expectation that monkeys will do what they were seen to do in this experiment. I don't think we could justify saying they would certainly act that way. If we could, we would assign P(E|H) a value of 1.0, but since we can't, we need a lower value. Since the hypothesis asserts a similarity with humans, we could get a number by figuring out the probability that a human, placed in a similar situation, would act the same way. I don't happen to know of any hard data on that topic, so I'll have to guess, and I don't think 0.7 is an unreasonable guess. In other words, I'm estimating that at least 7 out of 10 people would skip a meal in order to prevent someone else from suffering. And then I'm supposing that, if monkeys experience the same motivation that humans experience in such a situation, then it is similarly probable that they will act the way humans do in that situation. This gives us P(E|H) = 0.7.

As for P(E|~H), here we're asking: How probable is it that this evidence would exist if the hypothesis were false? If monkeys don't experience anything relevantly like human empathy, then how likely is it that, notwithstanding their lack of empathy, they would act as if they did have empathy?

It is not immediately obvious how we might estimate this probability, but I think I can suggest a way. Those who deny the hypothesis are not just talking about this particular species of monkey. Their claim is that empathy is unique to humans, that no other animal experiences anything like it, even if some of them, on some occasions, act as if they do?

Let's call this behavior, this acting-as-if, a simulation. Then we're asking for the probability that these monkeys would simulate empathy. And we could start by asking for the probability of any nonhuman species doing it. The claim is that only humans have empathy, and so if any nonhuman species seems to exhibit it, it's only a simulation. So then, in how many nonhuman species do we observe this simulation?

Practically none, as far as I know, but it might be objected that it wouldn't be fair to sample all nonhuman species. After all, nobody is about to claim that insects have empathy, or fish, or reptiles, or even most mammals. So let's shrink the reference class as much as we can. Shall we just check the primates? How many primate species simulate empathy?

Aside from these experiments, I don't know if anybody has even tried to look, but if it were common, I think it would have been noticed even without a deliberate search for it. This experiment, when reported, seemed to really surprise a lot of people, and so pending further discoveries, I think it reasonable to assume that such behavior rarely occurs among animals that don't really experience empathy. This gives us a P(E|~H) considerably below 0.5. Shall we say 0.2?

And now we can run the numbers, and we get P(H|E) = .78.

Those who think that figure is way too high have some options for counterarguments, but there are some mathematical constraints. Of course, if the counterargument is, "Bayes theorem is just irrelevant to questions of this sort," then the mathematical constrains don't apply. For the time being, I address my remarks to those who say, "OK, maybe you can use Bayes, but you must be using it wrong."

But the only way to use it wrong is to use wrong estimates for the three variables. Let's see how the output changes as we tweak those estimates.

Here is the first mathematical constraint: No matter what prior you use, if P(E|H) = P(E|~H), then the consequent probability of the hypothesis is equal to its prior probability: P(H|E) = P(H). In other words, if the evidence is just as likely to have obtained whether the hypothesis is true or false, then it is epistemically irrelevant to the hypothesis. It does nothing to either confirm or disconfirm it.

Next constraint: If P(E|H) > P(E|~H), then P(H|E) > P(H), and the reverse inequality also holds. That is to say, whether the consequent probability is greater or less than the prior depends strictly on whether the evidence is more or less likely on the assumed truth of falsity of the hypothesis. Further, the amount of increase or decrease between prior and consequent depends on the degree of difference between P(E|H) and P(E|~H).

Next constraint: If a prior probability of either 0.0 or 1.0 is assigned to the hypothesis, then evidence becomes irrelevant. For P(H) = 1.0, P(H|E) = 1.0 regardless of any probabilities assigned to the evidence, and likewise, for P(H) = 0.0, P(H|E) = 0.0 in all cases. For any other prior, the difference between P(H) and P(H|E) depends, as already noted, on the difference between P(E|H) and P(E|~H).

Whatever the correct values, I think it apparent that the observed behavior must be judged more likely if monkeys really do feel empathy than if they don’t, which means P(E|H) > P(E|~H), by some margin. And for this reason, the consequent probability of the hypothesis must be higher than whatever prior probability we give it, thus P(H|E) > P(H). And, for reasons I have given, I think P(H) = 0.5 is a very low estimate. This results in a consequent probability greater than 0.5, meaning that given the evidence, the hypothesis is more likely true than not.

But suppose you think my P(H) is still too high? In that case, whether the consequent probability exceeds 0.5 depends on how much higher P(E|H) is than P(E|~H). For any difference between P(E|H) and P(E|~H), you can find a value for P(H) that results in P(H|E) < 0.5, but you need to justify that value.

So, I have explained why I think my estimates are reasonable. If you think they’re unreasonable, we can discuss your reasons.

And if I’ve left some of your questions unanswered, let me know and I’ll try to get back to them.

If you'd like, feel happy to start by merely listing prior probabilities, and indicating whether they span the space of possible explanations.

You mean, if my probabilities are not sufficiently trustworthy, then I should think that anything not impossible is probable?

I'm not sure; I meant what I said to be qualified by the following sentences:

LB: Before you get [sufficiently trustworthy probabilities], it seems that intuition has to do the deciding. So, I'm trying to figure out whether you can do probabilities, or whether it's really intuition at this stage in the game

We could discuss what it looks like to have probabilities between a completely uninformative prior and probabilities with sufficiently compact distributions.

I'll try to include, in my exposition, some remarks about the role of intuition in a Bayesian analysis.

I would definitely appreciate that. I am very interested in the interface of fuzzy, intuitive thinking, and precise, formalized thinking.