I appreciate this response quite a bit! It gave me another chance to think through these issues with the changed perspective of an added year of reading. I’ll try to address the letter chronologically, as I read along with it, and as I do so, you’ll see I get carried away with certain themes. I only hope I haven’t missed your forest for the trees.
Note with stotting—I always see it discussed as a signal to the predator, but it also functions as a signal to fellow antelope. What gives?
It's a good point here that anti-inductivity, broadly, is the reason you need certain classes of costly signaling. Solutions to problems, if visibly successful, will be imitated, leading to solution fads (widespread adoption). Because the fitness of this solution is frequency-dependent, widespread adoption reduces its efficacy as solution. When the "point" of the signal—the signal's function—is distinction from one's fellow organisms, a comparative “I’m fitter than they are, so hunt them,” frequency-dependence is always in play. “If everyone’s super, nobody is.” Syndrome could’ve been the hero of that movie, provided another set of writers.
But what Syndrome really represents is inevitability. That which lends advantage and can be copied is copied. As Bourdieu convincingly argues, symbolic capital works by differentiation, or distinction. That which signals wealth or sophistication is inevitably pioneered by the upper classes and imitated by the middle classes, because there is advantage in it. (Accompanied by some countersignaling—the high appropriate the low.) Tyrion purple worked because it was literally impossible to afford . Now (perhaps) we have luxury beliefs and risk tolerance (social risk and otherwise).
And so, like you note, signals (which are surrogates—they are "not the thing" but "stand for the thing") are context-sensitive; their efficacy or costliness is environmentally dependent. You end up with “evolutionary holdovers” that are either “no longer so hard to fake” or “no longer strong indicators of the relevant outcomes.” No surrogate measure or metric is perfectly robust to environmental change, because any correlation or heuristic can be uncoupled under the right circumstances.
I want to linger a bit on the way that these evolutionary holdovers “can continue to have power, in so far as things that are desirable themselves become symbols of status, in a self-reinforcing feedback loop.” I tried to think of examples of vestigial signals that had no effect on reproductive fitness, and I couldn't. And the main reason was more or less what you say: even if wide birthing hips no longer give a birthing/survival advantage to women moderately affluent countries, the sheer fact that such women are perceived as attractive means that, if a man reproduces with a woman with wide hips, his daughters will have wide(r) hips, and thus a greater chance of reproducing. It's "speculative" and "extrinsic"—it's a bet based on other people's bets, a Keynesian beauty contest. Sexual fitness, separate from survival fitness. Although it's easy to imagine that reproductively fit individuals are better protected and more prized by their community, and therefore survive at higher rates, so "reproducing" and "surviving" don't uncouple cleanly.
And with sexual fitness, it comes down to desire in a similar way that non-sexual fitness does. In the same way an organism who looks unappealing to a predator (separate from the reality of his caloric benefit) will survive at higher rates, an organism who looks appealing to the opposite sex will reproduce at higher rates. Beauty is a two-player game.
Not every desire is adaptive, of course; our sensibilities often lead us astray. But that’s what desire is designed for, that’s its purpose: to guide us toward what’s typically good for us.
Desire says “is this good for me?” And beauty says “get over here! You know you want to!” Desire is a lock, and beauty its key. Desire chases beauty, and beauty chases desire.
Beauty is the result of an on-going selection game, where the desiring player selects according to what he believes furthers his ends, and the organism selected passes on genes. Beauty is a surrogate for fitness, like limping is a surrogate for "easy prey to catch," and stotting is a surrogate for "hard prey to catch."
We're in agreement thus far, but it's time for some conflict!
The first explanation we might instinctively provide for this difference is that stotting is not a conscious, calculated choice by the prey (“if I leap now, then that wolf will infer that I am strong, and realize how foolish it would be to chase me”), but rather behaviour that is essentially automatic, like some part of the fight or (in this case) flight response. That seems reasonable, except that presumably the “reading” by the predator is just as similarly unconscious and automatic (if that is indeed a fair characterization of the prey’s behaviour).
I wanna make two arguments against this. The first is that we have no idea what goes on in the mind of the deer. IMO, assuming that the deer is not making conscious decisions is as unscientific as "anthropomorphizing" it and assuming it is. Maybe less scientific, because anthropomorphizing is predicated on our evolutionary similarity and shared descent?
The second is that I'm not sure it's all that useful or coherent a carving to define writing by either consciousness or intent. There are extreme ends at which I think this is true: I think an herbivore trail, or the ethological concept of a cue, is “writing”—information is added to the environment through an organism's actions—but it is writing which is done as a byproduct of the primary action of the organism (e.g. browsing vegetation). When we walk dusty roads or snowy trails, we leave traces—we write evidence of our passage into the ground. When we log on to Facebook, we write that we are online, through our “online” marker. Sometimes we do this on purpose: we make ourselves online hoping someone will chat us, or set our status to “away” in order to have an excuse for not responding. Much of the time, the writing happens naturally in the process of logging on and logging off. But even then, we are often aware that we are writing information to others minds, and so we seek to limit or alter that writing. In my mind, it's incredibly hard to draw distinctions between these activities, given that I believe: (1) that following Trivers, our mind is constantly strategically acting in ways below our conscious awareness. (2) That we are constantly aware of both the informational effects (those that depend on the interpretation of reading organisms) and intrinsic effects (those that don't) of our actions, and consideration of both kinds of effects alters our acts. (3) That we often do not know exactly what we intend, determining our own motivations is very difficult, separating what is on accident from on purpose, etc. In other words, “consciousness” to me feels more like “was this made available at one level of an organism's control interface” than “was this action strategically modulated with reference to its informational effects.” Because so long as our projections of those informational effects alters our behavior—we don't wear that dress, we don't hang out in a neighborhood near the office on days we called in sick—then we are in some deeply meaningful way altering and modulating our "writing." And because this "writing" is impossible to escape—even our absence, our disappearance, alters the information state of others. And what to do when our writing does not have the effects we intended but does actually yield true information for the reader—like when our attempt at bluffing strength betrays our actual vulnerabilities? (You allude to this later, so I'll come back to it in a bit.)
It obviously has not carried out a precise statistical inference, but rather has applied various heuristics to do a kind of approximate inference, incorporating the new information into its model of each deer’s ability and it’s likelihood of catching each of them.
A year ago, I’d ++ you for this—having recently Gigerenzer & Taleb’s arguments, say, for the way baseball players seem to use simple heuristics, instead of unconscious calculus, to catch fly balls—but lately, skepticism waxes. My ideas on this are still incredibly crude—I apologize, but bear with me. It seems to me that our bodies are so incredibly precise with their movements, and our brains so remarkably good at inference from minuscule amounts of information, that it is difficult to believe it’s all "obviously" simple heuristics, instead of Bayesian-approximate statistical inference. At least at the lower levels of cognitive tasks, the kinds which our conscious mind delegates out. Second, even with all our novel inventions, visual stimuli, etc, our visual systems seem to only rarely fail us even when applied to subjects or domains that are unthinkable in the ancestral environment. My hunch is that if it were all simple heuristics, the rate of technological and social change we see, and the immense differences between ancestral and contemporary life, would lead to serious uncoupling.
That said, it is true that non-human animals appear to be less context-sensitive, less adaptable, etc than humans. They appear to be more instinctive, less acculturated. More hardwired, less plasticky. Which is to your point about humans having more flexibility to craft our signals or carry out our readings over an extended period of time. “Pragmatics,” in the linguistic sense of the “contextual modulation of meaning,” is—in my understanding—a key aspect of human linguistics specifically, and not so much animal communication. So, you may be right in the specific case, even if humans are less “homo heuristicus” than some anti-Bayesian Brainers would have it. (Although more homo heuristics than many pro-Bayesian Brainers hold—Compatibilism! Both sides are right!)
Especially on the writing side, it is no doubt the case that many times we very deliberately carry out actions that we believe will cause a reader to make particular inferences, and in some cases the calculated nature of this might sometimes lead this to be labeled manipulation.
I agree with the first separation, that sometimes we write deliberately, and sometimes writing is an inevitable byproduct of an action (or at least, of performing the action efficiently/cheaply). I take issue with the second claim, however: I believe that any time we deliberately carry out an action in order to cause a reader to make a particular inference, we "just are" manipulating them. Our representations of the world are fundamentally actions whose purpose is to bring about effects, a delta in the state of the world. The only way representations ("writing") can have such effects is by using a perceiver-interpreter ("reader") as its medium. The point of altering perception, then, is to alter action—full stop. It may not feel "manipulative" because it isn’t exploitative—it might advantage both parties; the information might be “true”; etc. But there’s never, in my opinion, a situation where deliberate representations are not intended to alter perceiver behavior.
Onto actual reading and writing, instead of our generalized version!
At the same time, it seems implausible that writers begin purely with such calculated modeling. Rather, they begin. Either through practice, raw talent, or experience gained through feedback, they have some intuitive sense of what will work, and follow those intuitions, or even begin in the most non-deliberate writing they can, hoping that things will emerge, or that they will be able to shape the material into something later.
The question is, “what will work [to accomplish what goal]?” And given that the mechanism by which the writer figures out “what works” is “what people positively respond to,” I think it’s fair to say that writing is “about” and oriented to getting a set of positive effects out of readers—praise, regard, certain responses like “this text changed how I view the world,” etc. Now, to the writer, it feels like they want to produce text that actually changes how people view the world. Perhaps this is their actual, ulterior motivation, and it just so happens that the only real evidence of this transformative quality is the self-report of reader. Perhaps it's positive response which they actually, deep-down care about. We can't really tease apart the difference, can we? Either way it's opticratic. Surrogate or not, feedback is the selection pressure which determines what kinds of things the writer writes. It isn’t calculated, in the sense of “pre-planned”—it just evolves in the direction of its reward function, like we all do, always. That’s what progress is. That's what "good" is, more or less.
But you're exactly right that one effective strategy is to generate a lot of options, then keep what’s good. (As in automatic writing.)
Which of course is manipulation, like the insertion of music in a movie scene to “trigger emotional cues.” It’s not that the critic dislikes being manipulated: he loves it when it's done well. He just doesn’t like clumsy, transparent, artless manipulation (like the rest of us). And plus, the art encounter is a low-stakes "sandbox"—we might weep when the main character dies, but that's about as far as real-world consequences of manipulation go. Whereas in the real world, being under another's control can cost you everything.
(I also like this distinction of automatic reading vs considered, deliberated reading. The latter is far more context-sensitive and thus “human.”)
On Michael & Knapps, I’ve become a bit more sympathetic, since 2020, to their intentionalism argument in light of my Schutz reading—Schutz thinks that the normal interpretive mode of human beings, which he calls “verstehen,” is one which understands human actions, representation, etc primarily in terms of the actor’s “project”—the set of goals they’re trying to accomplish. “The task of fellow-actors, therefore, is necessarily one of inferring from a fragment of the other's conduct and its context what the other's project is, or is likely to be.” But I think that this is true because a person’s project in conjunction with theory of mind—that is, their algorithm of desire as filtered through belief, desire as made achievable or blocked by perceived affordances and constraints—is the most compressively powerful way of modeling them. Your crossword example illustrates this pretty well actually:
The goal of the puzzle solver is largely to infer what the writer intended for each clue, although in rare circumstances, there could be multiple solutions which would fit equally well, in which case one might fail to infer the intended meaning, and yet still “solve” the puzzle.
Intent here is a surrogate for a working solution to the problem.
Similarly, when we ask, is the “meaning” of a text the intention of the author, or the interpretations of the reader, our answer depends a lot on the perspective one takes on the system. If you’re taking the perspective of a text + world, and model the text as a stimulus, then all reader interpretations are part of its total effects. But if you take the perspective of a given reader, you are more or less trying to figure out the intent of the writer, because that intent unlocks the organizing principle, like it determines the solution to a crossword—the logic by which all the parts fit together. That logic is also what the reader can learn from the writer, insofar as the text has been vouched for, or is produced by a celebrated mind, then the writer comes for the insights of that mind or text. In cases where a text has been lauded for a certain message or set of insights which were not intended by the author, the reader’s probably best off interpreting it through that lens first! But it's about the goals of analysis—entailment for world, or entailment for you the reader.
This gets back to the example of a speaker attempting to bluff strength but betraying to the listener their real weakness. This is a core problem in hermeneutics: that there is information intended by the author, and then information which is yielded by a hermeneutics of suspicion—what does the way the author described a character tell us about their culture, their beliefs, their views? Which I think can only be answered by getting into the weeds of what information is: a difference that makes a difference. An indicator which inference is performed on. All different kinds of understandings, some close to the truth, some far from it, can be drawn from these indicators. The author at most intends a few of those inferences. If the meaning of a text is “all true inferences drawn from its information”—all [worldy] differences that are testified to by by its [textual] difference—then the space of intentionality is minuscule in comparison with all the true inferences that can be pulled out of a text.
Which I now see you get at perfectly, so I'm more or less in preemptive recap mode:
Moreover, there is simply much more than can be inferred. Indeed, much of the elaboration of criticism is in thinking about art as information which can be used to make inferences about the world, either about the world in which the work was created, or about aspects of the author which they themselves may not even be aware of, at least not consciously.
I do think you’re right that, because we can’t do continuous inference over many domains, particularly domains that are constantly changing up on us, so we use heuristics, the biggest of which is typification—reading parts to make a guess about whole—the superset of the politically charged “stereotype,” basically. Perhaps this contradicts my earlier arguments regarding the limits of “homo heuristicus.”
