The title is from a blog post by Brian Leiter. Leiter links to an article in the LA Review of books: Imitation and Extinction: The Case Against Reality. The article is written by Donald Hoffman.
We have discussed the general topic before, in several threads. So maybe this is a good time to revisit the topic.
Hoffman asks: “I see a green pear. Does the shape and color that I experience match the true shape and color of the real pear?”
My take is that there is no such thing as the “true shape and color of the pear.”
It is a common presumption, that there is an external standard of truth. Here, I mean “external to humans”. Truth is presumed to come from somewhere else. And our perceptual systems evolved to present us with what is true.
As I see it, this is backwards. Yes, our perceptions are mostly true. But this is not because perception is based on truth. Rather, it is because our human ideas of truth are based on what we perceive.
Open for discussion.
I just read the LARB article, and it’s not clear to me what Hoffman even means by “veridical”, since he isn’t a realist:
If there is no independent reality, then there is nothing relative to which perceptions can be veridical (or not).
Hoffman then seems to contradict himself with this:
What we have is language and other behaviors that are reinforced, or not.
I learn what color means from interacting with the language community. Insects learn what color means by attempting to find nectar. The meaning of reality is embodied in interactions.,
Of course there is an independent reality (in my opinion).
But so what? There is still nothing relative to which perceptions can be veridical.
People have conflated reality with truth. But they are not the same. There can be an independent reality without there being an independent truth.
Is anyone familiar with how he uses Evolutionary Game Theory and fitness payoffs? He claims simulations based on this theory show that fitness payoffs mean evolutionary success does not imply that successful organisms have developed sensorimotor systems which latch onto the causal structure of reality relevant to surviving in their niche.
I saw nothing on perception in the SEP or Wiki articles on EGT .
Otherwise, there is a lot there that we have covered at TSZ: scientific realism, nature of perception, primary versus secondary properties (shape versus color), illusionism ( he re-uses Dennett’s icons on a desktop), nature of truth, mental representation and its relation to reality, EAAN, evolutionary niches.
The bit on the world being fundamentality consciousness (or the shared virtual reality of conscious agents?) sounds like some new age version of idealism. I’ll leave that for JMac.
ETA: Wiki article on Donald Hoffman links to a pdf of a paper he wrote defending his consciousness-first theory. Perhaps he explains how it relates to idealism there. Also how he can draw conclusions about reality from evolution when science is based on illusions. But life is too short for me to spend time with that paper. For one thing, I may get a new reading assignment from Gregory.
Neil,
Sure there is. To alter Hoffman’s example slightly, suppose there is a plastic grape on the table. One person sees it as a food item. The other sees it as a fake. The second person’s perception of reality is more accurate — more veridical — than the first’s. In reality, there is no food there.
Reality is what there is and human perception is more or less accurate (and the scientific method an attempt at improving accuracy)~. “Veridical” is a legal concept.
Alan,
No, and we’ve had this conversation before. Back then you thought “veridical” applied only to speech.
Not so. “Veridical” applies to perception too.
Note definition #2 below — particularly the usage example:
I am failing to see the relevance.
Hoffman’s concern was with evolution. Unless plastic grapes are common in the environment, the example does not seem to apply.
The plastic grapes are just one concrete example. There are plenty of evolutionary scenarios in which organisms need to distinguish between food and non-food, or between danger and non-danger. All quite relevant to evolution.
But in any case, we were discussing your claim:
Perceptions can be more or less veridical relative to that independent reality, as my example shows. It matters whether the food you perceive (or don’t perceive) is really there, and it matters whether the danger you perceive (or don’t perceive) is there too.
Bruce,
The following extract, taken from here, gives some indication of what he means:
He seems to be making the rather obvious point that some features of the environment are more relevant to specific organisms versus others, and at specific times versus others, and that perceptual systems are tuned by selective pressures to focus more on the relevant features and less on the irrelevant ones.
But who ever doubted that?
keiths,
You completely miss the point. But I’ll leave it at that.
Neil,
I doubt that.
Don’t give up so easily. See if you can actually defend your position.
Way to miss a point. I’m referring to the concept not the dictionary. Perception is never veridical. See what petrushka is saying.
Right. And there are many articles by he and his colleagues at his web site on his observer theory of perception.
https://www.cogsci.uci.edu/~ddhoff/
But is it just a pet theory that few others in the sciences of perception research pay attention to? I suspect the answer is yes.
Hoffman has made appearances in several podcasts and has written for other pop sciences sites.
https://www.quantamagazine.org/the-evolutionary-argument-against-reality-20160421/
He also has an new general audience book out. Definitely a correlation there.
There is a review article on this theory with various responses here:
https://journals.sagepub.com/doi/10.1068/p4302ed
The issue seems to be open access. I’ve glanced through many of the articles. They seem more philosophy than science, but the Anderson one does question the scientific assumptions and generalizability of Hoffman’s simulations.
ETA: The Cohen article linked in his comment on the Leiter blog is in this issue. He closely examines the meaning of veridical and critiques the assumptions of the simulations using that examination.
https://link.springer.com/article/10.3758/s13423-014-0782-3
I find his criticism convincing.
The Leiter blog post has been updated with a comment from Jonathan Cohen. Leither has opened up the comments section. Here is Cohen’s comment:
“just saw that you posted a link to Hoffman’s latest repackaging of the no-selection-for-veridical-perception thing he has been circulating for years. I would have left a curmudgeonly comment but you (sensibly) left comments off the post.
but just to say: Hoffman’s piece is seriously confused about some basic philosophical issues concerning the nature of representation and veridicality, and his confusions completely undermine his arguments — including the evolutionary game theoretic argument, which just build them in.
so, for example, his attack on representation depends on construing that notion in an implausibly strong way (for A to represent B, it must be true that A=B) that, as far as I can tell, isn’t defended by any serious proponent of a representational construal of perception.
another example: while the headline claim is that selection favors an interface strategy tuned to payoffs over a strategy that veridically represents resources, what the simulation actually shows is that one strategy involving veridical representation (of payoffs) beats another strategy involving veridical representation (of something that the game is constructed to make come apart from payoffs, viz., resources). that hardly shows that selection favors strategies that avoid veridical representation.
a number of people have tried to point out some of the problems, but Hoffman hasn’t seemed to display a lot of uptake over the years. ah, well. (fwiw, my own short response to him from the Psychonomic Bulletin on the topic in 2015 makes some of what I think are the relevant criticisms in a reasonably readable format: https://link.springer.com/article/10.3758/s13423-014-0782-3 .)
and now that that’s off my chest I will descend from the soapbox….”
The standard corresponding example in philosophy is small black things and frogs, specifically, in what sense has the frog misrepresented a small black thing as food when the frog flicks its tongue at it in experiments.
https://plato.stanford.edu/entries/content-teleological/#3.2
The challenge is to formulate a theory of misrepresentation which is not circular. That is, which does not rely on a homunculus which decides in the frog’s brain/mind or on the judgement of a scientist or forum poster observing the experiment.
The standard philosophical answers rely on some combination of fitness in evolution and contribution to frog’s homeostasis and ability to contribute its gene’s to next generation. That’s one of the oft-discussed-at-TSZ topics I was referring to in my first post.
Thankyou Neil for bringing the work of Donald Hoffman to my attention.
In Saving the Appearances Owen Barfield has long since discussed the idea of Donald Hoffman that we humans can come to an agreement on entities that we perceive. Barfield called this shared agreement “collective representation” But unlike Hoffman who posits a hidden reality behind the world of our experiences Barfield argues that our experiential processes can give us access to reality, Hence the title of his book, “Saving the Appearances”.
Indeed. And it seems to disagree with what keiths is arguing.
Another problem, here, is that the meaning of “perception is veridical” is far from clear.
John Searle, in his book on perception, just uses “veridical” as contrasted with delusional. So if the frog cannot distinguish non-flies from flies, that would not say anything about whether its perception is veridical. But if the frog is just dreaming of flies, then that is not veridical.
To me, this is a weird way of talking. Having dreams or delusions, is not perception at all. But Searle disagrees, and argues against disjunctivism (the idea that delusions and dreams are not perceptions at all).
This is what I take keiths to be doing in his argument.
That’s the wrong link, although the linked article is interesting. Here is the link I meant:
https://link.springer.com/journal/13423/22/6/page/1
The Cohen article explains what Hoffman means by it for those of us not up to reading the Hoffman article [ETA] in the journal issue in which the Cohen article appears.
Confirming what others have already said, I think we should definitely distinguish the question “is perception veridical?” from “does Hoffman’s simulation show that perception is not veridical?”
One can think that the answer to both questions is “no”.
Searle is a textbook example of a philosopher who thinks that one doesn’t need to know anything about science or about any other philosophers in order to feel entitled to an opinion — and an audience.
Also, Berkeley stripped him of his emeritus status because was found guilty of sexually harassing one of his grad students (see here).
The way this is presented seems to suggest that if not for veridical perception, then it selects for false perceptions, which puts it in very artificial, equivocal, terms.
Evolution selects for good enough, and if it gets slightly not as good, or slightly better, so be it. This is why there can be an arms race between deceivers and perceivers/interpreters. So, it’s not that perceptions are “veridical” or “false,” the perceptions are responses to some stimuli (I’d say always “veridical”), and they vary in degrees of “sophistication” depending on the challenges in the environment, which can contain things that might fool interpretations of what’s perceived. If those things are abundant enough to challenge survival, then more refined perception and interpretation organs/systems might evolve.
Putting it in terms of perceptions themselves being “veridical” makes the discussion a bit muddy.
keiths:
Alan:
The word refers to the concept, and both are applicable to perception. It’s standard usage.
Click here.
ETA: Ditto for “non-veridical”.
Entropy,
Right. And his notion of veridicality also seems way too narrow.
keiths,
Maybe. However, it works very well when I try and figure out what’s going wrong with some “arguments.”
I think there are lots of difficulties with the very notion of “veridicality” here.
Supposed we were to accept provisionally that natural selection will tend to select against non-satisficing perception of salient motivating affordances.
Would that count as “veridical perception” or as “non-veridical perception”? Why or why not?
Did you mean “satisficing” and “select for”?
I has a small conversation once with one of those presuppositional imbeciles and the poor idiot insisted that we know that the sky is really blue because “God” made it blue for us. Leaving aside the physics, and if what we really see is light dispersion, or the fact that it only looks “sky” blue during such and such times, red, orange, yellow, dark blue, and even black at other times, I told him that the color is “really” blue because we were the ones who decided to call that color not just “blue” but “sky blue” ourselves.
Anyway.
I was trying to think of natural selection as the elimination of traits that fail to satisfice.
Without getting into the philosophy (which I am utterly incompetent to discuss), let me make one point. Let us assume, for the sake of discussion, that there is a real world out there. In that case, the fitness of a phenotype is not a matter of individual opinion, but something real. If a bird sees grapes and spends time trying to get at them, this would either be good or bad for its fitness depending on whether or not the grapes are real grapes. In which case natural selection would tend to select for kind-of-real perception (“viridical” perception?).
Of course there would be tradeoffs with how much time and energy had to go into that perception, so we would end up with nervous systems that are capable of being fooled by optical illusions. Those would mostly be rare in nature, except for mimicry and camouflage, which are themselves the result of natural selection.
Umm….
What hit’s our retina is light of various frequencies. We see many things.
We see colors in dreams, when no light is hitting our retina. We see mixtures of frequencies as something different from any one taken individually. We see afterimages. We see things when hit in the head. We see colors in a Benham top.
Interestingly, brain damage can eliminate the ability to see color, to dream about color, to imagine or remember color, to see color illusions.
Color is the poster child for perception as distinct from input.
Oh the irony! 😱
Joe Felsenstein,
Perception just needs to be adequate for the niche. Veridical Schmizical.
Alan,
That quote is not from me.
I’m not on board with “satisficing” or with “adequate to survive”. Because if a change of genotypes could do a bit better than satisficing, that would have a fitness advantage, and the fact that the previous genotype satisficed would not preserve it from being replaced.
Instead I suggest that mediocre performance can result from conflicting selection pressures when there are tradeoffs. That’s a very different notion, and does not involve any calculation of “good enough”, but rather consideration of whether the further gain in performance is worth the loss in fitness from expenditure of resources on that aspect of the phnotype.
Anyway this is a separate issue from veridical cognition.
Here’s the key paragraph from the Jonathan Cohen quote:
The fitness of the phenotype is what corresponds to the “payoffs”, so the changes in the population should correlate better with it than with other measures. Which is an admittedly-after-the-fact rationale for their result.
Then we could also say natural selection doesn’t select for tall. It doesn’t select for short. It doesn’t select for smart. It doesn’t select for dumb. Natural selection doesn’t select for fast. It doesn’t select for slow. Natural selection doesn’t select for strong. It doesn’t select for weak. Natural selection doesn’t select for big penises, it doesn’t select for small penises. Natural selection doesn’t select for straight teeth, it doesn’t select for crooked teeth. Natural selection doesn’t select for white, it doesn’t select for red. Natural selection doesn’t select for good singers, it doesn’t select for bad singers. It doesn’t select for deep voices, it doesn’t select for high voices…
Why would veridical perception be unique?
What an important sentence this is. This applies across the board to the ENTIRE premise of natural selection. This would ALWAYS be the case. Its why you could never start off building a good car, by building a light bulb first, then slowly changing that light bulb into a light bulb that washes clothes, and then slowly taking that light bulb that washes clothes and putting wheels on it so it can be driven…
If we are always selecting for 10,000 different traits at the same time, I would call the results only mediocre, I would call them cartoonish.
Its the most obvious flaw in the entire concept of natural selection as a designer. But you will be very hard pressed to find many skeptics who can admit this.
Cue the niche moron, who thinks niches are discreet and enduring, and non-overlapping. And have only one solution.
@phoodoo:
Accurate perception, unlike all those other characters, would tend to be associated with gains in fitness. But there are tradeoffs, so like components of fitness it wouldn’t get infinitely extreme.
Tail lengths, size, straightness of teeth, etc. probably have values that are optimal (given the environment the other characters).
Biologists have to think about these tradeoffs all the time, and even measure them by looking at the relationship of those characters to fitness and their genetic correlations with each other. Human designers also have to work with tradeoffs all the time.
This would of course apply to animal and plant breeders as well, but somehow they seem to have made progress. I believe that is an analogy Charles Darwin made early in his book.
I wrote above:
I should clarify. I meant that there are values that are optimal (in a given situation), not that the species necessarily has those phenotypic values.
Why not, still not enough time?
Joe Felsenstein,
Joe: If you are interested in spending the time on it, do you have any thoughts on the Evolutionary Game Theory modelling that Hoffman does in eg this paper:
https://link.springer.com/article/10.3758/s13423-015-0890-8
ETA: Cohen mainly criticizes the way Hoffmnan sets up his model of perception using measure theory. But Cohen does not address the EGT work itself — there is no need if the setup is wrong.
But I was wondering if you had any thoughts on the EGT work itself, even supposing the setup was reasonable.
A few more comments at Leiter blog which support Cohen:
https://leiterreports.typepad.com/blog/2019/09/evolution-does-not-select-for-veridical-perception.html#more
I find this comment to be very funny. Do you think breeders are selecting for 10,000 different traits at the same time or one trait? What would be the result, if you were trying to breed for a Siberian Husky, and you did it by selecting for the traits for ten different dog types at the same time?
I’ll say. It is completely unclear to me what is the question of the OP, and the paper it refers to.
When is a perception non-veridical? If a frog perceives a small black dot when there really is an small dark object in its field of vision, that seems like a veridical representation to me. But if it then mistakes that object for a food item, that same perception is not veridical? I don’t understand that.
If we bring fitness into the definition, then a lot of human perception becomes completely irrelevant. For example, a considerable portion of the human population have red-green colour blindness. The impact on their fitness is pretty much zero. Yet, I would not say that perceiving redness is a non-veridical aspect of (say) a piece of fruit.