On science, beauty and utility

I have a hard time defining beauty & art-- I think everyone does. To wit-- I think there is probably a factor in inspiration; little models of atoms or cgi of ribosomes at work &c-- that bring people into the fold, or that make things "click." Pictures of space, of cells, of nature-- heck, Natural Geographic-- combine science & art pretty neatly.

I tend to think the word "art" is applied very broadly these days. When discussing visual art, I usually try to pin the label down a bit further, either by calling it visual, fine art, illustration or what have you.

It is legitimate to refer to non-artsy things as art, since the skill levels and sense of beauty often involved in the sciences or medicine deserve it.

Hmm. Stuff to ponder.

Makes me think of Bourdieu (the sociologist) who argues that aesthetic preferences are learned, and play a big part in shaping socio-economic subcultures. Beauty as a social construct. William James also has a nice story to that effect somewhere; After deploring that a poor community somewhere has cut down the trees of their immediate surroundings, he is surprised after a conversation with one such community member that they actually find their new surroundings much prettier.. Ah well, the shit you see in shops and houses..

Also, shouldn't aesthetic judgments be reserved for human artifacts? Nature is always beautiful (although I know a lot of people who are hard pressed to admit that spiders actually are very pretty)!

Why do you suppose he cited the NSF but not the NIH budget?

Part of the pleasure of science and art comes from seeing a beautiful pattern, and that's only one place where the two disciplines come together. Not to oversimplify, but don't art and science both create(or find)order from disorder?

Love this post.

Excellent post, BioE!

You may well argue that vaccine development is more important (and more worthy of funding) than viral capsid aesthetics.

There is also a very good argument that the study of viral capsid aesthetics is at least as important for eventually achieving effective vaccines as is research with the explicit goal of vaccine development.

It seems to me that much of the difficulty that surrounds issues like the "public understanding of science," science funding, grant writing, and the like stems from the fact that most (or most of the best) scientists are mainly motivated by broadly aesthetic concerns (not just the beauty of nature, but also the joy of exploration and discovery), whereas funding is mostly driven by purely utilitarian or even narrowly economic considerations. This is why the research that gets funded is often not the research that the scientists most want to be funded, why grant applications can be so painful to write, and why press releases about even the most intrinsically fascinating new findings almost always seem to feel the need to tack on some transparently self-serving, and often quite implausible, remark about how this research "might eventually" lead to a cure for cancer, or cheap, green energy or whatever.

I don't have any solution to offer, but it seems to me that this particular aspect of the disconnect between the interests of scientists (and also many of those laypeople who have a positive interest in the sciences) and the interests of funders often fails to be properly acknowledged, perhaps because scientists so often seem to feel that they need to dissemble, in public, about their real (and, in some respects, their more honorable) motivations.

"Nature is always beautiful," Jan-Maarten says. And I agree. But I wonder why?

A sexually potent 20-something is beautiful for a reason, to attract an equally potent mate and produce competitive offspring. That kind of Darwinian purpose I understand.

But what's the point of nature as a whole being beautiful? What Darwinian purpose is being served?

Is our attraction to the beauty of nature actually good for our survival? That we who appreciate nature's beauty out-compete, on balance, those who just commodify it?

Just asking. Great post.

I am an engineer/inventor, and I do my best work with my sense of aesthetics. I see aesthetics in art or science as a sort of pattern recognition, a mapping of one type of idea onto the framework of another type. I see them as very similar. Mapping of images or structures produced by a human and which invoke certain human emotional states, or mapping of physical structures/properties of one type/class of objects onto another type/class of objects.

When you do this mapping/transformation in your own neural network (a non-algorithmic computation), you are experiencing aesthetics. When you do it with an algorithm, you are doing science. In science, usually that mapping is to a mathematical model.

Human feelings and stuff are too complex to be represented with an algorithmic approach (because they are generated non-algorithmically in the corresponding neural networks), so science fails with respect to doing human-type stuff unless that stuff is trivially simple.

Beyond a certain level of complexity, algorithms become too complex and cumbersome to be able to be manipulated in useful ways. This is where a scientist has to have and be able to use their sense of aesthetics. If they have not developed a sense of aesthetics that closely corresponds to reality, they become lost in the details that are too numerous to fit into an algorithm.

A (good) scientist will not proceed into regions where ideas developed through aesthetics cannot be verified through algorithms. Non-scientists have no way to verify ideas with algorithms so they necessarily are only in such regions.

All pattern recognition is subject to Type One Errors (false positive) and Type Two Errors (false negative). If the sense of aesthetics does not correspond to reality, then one is doing magic/religion/delusion. The strongest way to test an idea against reality is via an algorithm. If you are unable to abandon the ideas produced via your aesthetic sense, then you are unable to do real science and are limited to doing what Feynman called cargo cult science. The ability to monitor and when necessary modify your sense of aesthetics so that it corresponds to reality is the essence of intellectual integrity that it takes to be a great scientist.

This does relate to my most recent blog, theory of mind vs. theory of reality. I see that as the fundamental trade-off along the autism spectrum, where people with ASDs have brains hard wired to be more "algorithmic", but that interferes with communicating with NTs who do their communication with the fundamentally non-algorithmic theory of mind.

This post is generating some outstanding comments! As mordicai and Glendon note, defining "beauty" and "art" is a problem unto itself. I read a book a couple of years ago on the social function of beauty, which never once defined beauty (and explicitly stated that such a definition was beyond the scope of the book). It was downright odd.

Jarett, I have often pondered the same question: it's puzzling why nature is universally seen as beautiful (except for things such as spiders and snakes which, as Jan-Maarten notes, arouse disgust in many), when the evolutionary advantage for that response is unclear in most situations. The two explanations I usually turn to are (1) it's the result of spandrels - finding certain things (mates, fruiting plants, etc) beautiful for evolutionary reasons bleeds over into a broader capability for appreciating beauty; and (2) I think EO Wilson is on to something with there being some evolutionary advantage to a more universal biophilia. Daedelus and zeladoniac, you also make excellent points about pattern-finding.

Nigel, I think your comment is probably right on. The question is, is it wrong for research funding to be driven by purely utilitarian concerns?

A (good) scientist will not proceed into regions where ideas developed through aesthetics cannot be verified through algorithms.

This is ridiculous. There is plenty of important and useful science that proceeds in the absence of algorithmic--i.e., completely deterministic mechanical--verification. Scientists justifiably believe many propositions that cannot be verified algorithmically.

The question is, is it wrong for research funding to be driven by purely utilitarian concerns?

As a practical matter, deciding what to fund also requires addressing the corollary question, Are purely utilitarian concerns best served by focusing funding on explicitly outcome-driven research?

"Scientists justifiably believe many propositions that cannot be verified algorithmically."

No. In so far that there is a belief that cannot be verified algorithmically, that belief is not a scientific belief. It may be a good hypothesis, it may be a reliable heuristic, it may be aesthetically pleasing, but it is not a scientific belief (which is itself an oxymoron).

If you dont have a way to algorithmically verify something, you dont have a scientific justification for believing it. There are many things that many scientists believe that have not been verified algorithmically. Some of them cant be verified because they are in fact wrong.

The problem with only funding utilitarian directed research is that we don't know how valuable something is until after we understand it. It is like doing library research where there are no titles and the utility of a particular book is only understood after reading it.

If you dont have a way to algorithmically verify something, you dont have a scientific justification for believing it.

Dude, that's ridiculous. Have you ever actually done any scientific experiments and interpreted the results of those experiments?

The problem with only funding utilitarian directed research is that we don't know how valuable something is until after we understand it. It is like doing library research where there are no titles and the utility of a particular book is only understood after reading it.

Exactly! Excellent analogy.

I have done many experiments and read about many experiment and interpreted the results of many experiments. None of those interpretations rise to the level of "belief" (as I understand how most people use the term), they are all tentative and I am prepared to abandon them if a better interpretation comes along.

Things that can be verified algorithmically are much more certain, such as 1+1=2. Those things can rise to the level of belief.

In the context of this discussion, conclusions arrived at solely via aesthetics are never certain enough to constitute scientific beliefs. Whether something is aesthetic or not tells us something about the neural networks that humans use to determine something to be "aesthetic"; it tells us nothing about the reality of that something.

I am only somewhat pulling your chain. Scientific reliability requires that every step along the way be subject to analysis, including the process of that analysis. Non-algorithmic calculation processes cannot be subject to analysis the way that algorithmic processes can. Aesthetics is ok for hypothesis generation, but the verification has to be done algorithmically.

"As a practical matter, deciding what to fund also requires addressing the corollary question, Are purely utilitarian concerns best served by focusing funding on explicitly outcome-driven research?"

Precisely. Outcome-driven research is most certainly insufficient to advance science as a whole. Funding only outcome-driven research projects may be more cost-effective in the short term, and more effective at meeting urgent needs, but such a strategy neglects the open-ended exploration required to develop new avenues of research for the long-term - avenues which will in turn deliver opportunities for outcome-driven research decades from now. A balance is required.

Dude, I think maybe we each mean something different by the term "belief". To me, "believing" something to be the case from the standpoint of scientific justification means that I am willing to move forward as if it is the case, designing and performing additional experiments and generally pushing ahead. This doesn't mean that I discount the possibility that the belief will turn out to have been erroneous.

If, as it now sounds, by "belief" you mean something that is not subject to later revision or rejection, then what you are talking about is not something relevant to the experimental sciences. There are *always* logically possible reasons why a particular experimental result might not imply what we think it does.

The search for certainty of belief in the way that you are talking leads only to the most mudane and boring conclusions--like 1 + 1 = 2. Most experimental scientists I know are more concerned with making headway on interesting empirical questions, and are willing to sacrifice the uninteresting kind of certainty you seem to require in order to do so.

Precisely. Outcome-driven research is most certainly insufficient to advance science as a whole. Funding only outcome-driven research projects may be more cost-effective in the short term, and more effective at meeting urgent needs, but such a strategy neglects the open-ended exploration required to develop new avenues of research for the long-term - avenues which will in turn deliver opportunities for outcome-driven research decades from now. A balance is required.

BioE for Director of NIH!!!

Love this discussion. I'm throwing Alberti's beauty definition into the mix:

"I shall define beauty to be a harmony of all parts, in whatsoever subject it appears, fitted together with such proportion and connection, that nothing could be added, diminished, or altered but for the worse."

Akin to gestalt theory, great beauty is achieved through a harmonious combination of parts, that together are more than the flat sum of their parts. And it exists in all fields of study and work. Even in utilitarian engineering then, as daedalus2u above admitted, beauty is integral to all gears/parts/circuits working together.

So physioprof, all these "scientific" ideas that you want me to believe in, how exactly do you know they are valid? Can you show me the chain of facts and logic that lead from observations to the idea? i.e the algorithm by which you have verified them to be correct?

Or do you want me to simply "trust" you because you "know" and "feel" that the ideas are correct, but you can't show me how you arrived at them?

Or are they simply valid enough for you? If that is the standard, then there is no standard.

Or do you want me to simply "trust" you because you "know" and "feel" that the ideas are correct, but you can't show me how you arrived at them?

HAHAHAHAHAH! No, dude. Of course we always explain in as much detail as possible the reasoning by which we arrive at a conclusion on the basis of experimental results. The point is that there are always an infinite number of mutually incompatible conclusions that can be drawn from any given collection of experimental results. Because theoretical conclusions are underdetermined by experimental results, one must always use a non-algorithmic process for deciding which conclusion one chooses out of that infinite number of possibilities. One example of such a non-algorithmic process is the invocation of Occam's razor.

The only alternative is to not draw any theoretical conclusions whatsoever from experiments one performs, and limit one's conclusions to a simple restatement of the experimental observations themselves. This would be both painfully boring, and counterproductive to scientific progress.

We never "know" with certainty that anything we believe about the nature of reality is valid, other than tautologies such as 1+1=2. We simply have no choice but to behave as though we do, but leaving open the possibility of being proven wrong in the future.

Let me try a different approach,

You seem to believe ideas because they are interesting, aesthetically satisfying, compelling and simple. I prefer to believe ideas that are correct. Perhaps an example showing the flaws behind your approach will illustrate the difference.

An idea I presume you are familiar with is the idea of homeostasis; is that a correct idea, a wrong idea or a not even wrong idea?

I prefer to believe ideas that are correct.

Sounds great! I heartily agree!

Please tell us how one can distinguish non-tautologous correct from incorrect ideas.