« Normal Ending: Last Tears (6/8) | Main | True Ending: Sacrificial Fire (7/8) »

February 04, 2009

Comments

Do people with Asperger's, who are insensitive to social signals, therefore have more free will than the rest of us?

And yet they tend to follow the same patterns day in, day out. Maybe all that repeating we observe is their subconscious replacing the missing social control with habitual control.

To what extent could the causation be going in the other direction? Perhaps the researchers have merely identified markers that are statistically likely to accompany competence. The competence would lead to success.

For instance, good instructors will tend to utilize every channel of communication, including non-lingual ones, even if those channels don't contribute the main impact of their lecture. Hence instructors using non-lingual channels would tend to be rated higher. The same reverse causation could easily be seen with good salesmen (they want to eek out every possible advantage, hence use every mode of communication even though the non-lingual ones may be substantially less helpful).

I made a little fun experiment a few days ago, somewhat relevant to this and to what Jordan says.

I got 20k photos with attractiveness ratings from one website, and tried some correlations between attractiveness percentile and photo properties. I got correlations like +0.2 on average pixel color (something like red + green - 2*blue), so I thought I found something quite significant and easy to manipulate. So I converted all pictures to grayscale and did pairwise testing of pictures with very similar original attractiveness and dissimilar original color balance - if color balance really made photos attractive, those that had identical attractiveness in spite of bad balance should be on top in grayscale most of the time.

And there was almost zero correlation. It seems that attractive women simply know how to make attractive photos, and unattractive women don't, just like Jordan says. It wasn't very rigorous experiment, I'd consider it a weak Bayesian evidence at best.

This blog post is pretty interesting, but I found it frustrating that the only link leads through a paywall, and there aren't any researchers' names or web addresses listed.

My guess is that Alex Petland is one of the researchers.

We really need better publication formats for good amateur science of the sort that Tomasz just described.

Eliezer: It's called blogging and it already works well enough, good blogs get a lot more readers and a lot more followup than most published papers. I would post it on my blog if I had some good results, but it's another case of publication bias against results supporting null hypothesis.

+1 for supposing Alex Petland is involved. See his book: Honest Signals: How They Shape Our World

Pedantic Note: @ Johnicholas and dzot: You mean Pentland, not Petland, and he's been known to go by "Sandy" rather than "Alex," or so I've heard.

@Alan - Thanks for catching my typo! I'm embarrassed.

@Tomasz Wegrzanowski - Please blog your experiment somewhere, I'm interested. I'm particularly interested in where you got your data, both the first set and the pairwise testing.

Thomasz, no. Eliezer is correct. Functional isn't close to optimal in this case.

Robin, great post.

After reading Pentland's book, I do not find that the high level summary quoted by Robin is supported by the actual experimental results.

@Forrest Bennett: could you explain this in a bit more of detail, I haven't read Pentland's book.

A question in brain anatomy:
How big are the newer brain structures with linguistic and deliberative reasoning capabilities in comparison to the older, primitive regions?

"A person's responses can often be explained by non-linguistic behaviours of other people and simple instincts for social display and response, without any recourse to conscious cognition.... Our conscious minds are more PR folks than CEOs of our total minds."

It's not entirely clear what reasoning is being offered here. Note that the mere fact that we can predict how people will respond to certain stimuli does not show that we lack "conscious control". (I can predict that most people will choose chocolate over dirt, for example, without this entailing anything about the role of conscious cognition in their decision-making.)

Instead, the argument needs to be that people's responses can be explained by factors that we wouldn't recognize as motivating our decisions (and that these factors are the causes, and not mere correlates of transparent motivations).

Anyway, I just thought that was worth making explicit, since it's a common fallacy (even if no-one here was really making this mistake) to think that prior causal influences somehow preclude our (conscious) selves also being causally responsible for our behaviour.

this is a really good post. thnx

Richard,
do you think that you're saying the same thing as Jordan?
(Maybe he's complaining that the researchers are overreaching, while you're complaining about the RH's jump from the reports?)

TAW,
blog it!
I would like to see scatterplots of score vs r+g-2b, r vs g, etc.

What do you mean by "attractive women simply know how to make attractive photos"? Do you mean "attractive women happen not to use blue"? How does this match Jordan?

It's certainly not the case that, say, blue is ugly and attractive women know not to wear blue; they would lose that advantage in grayscale. I see Jordan's hypothesis as being that color detects something (say, amount of skin) that is not destroyed by grayscale.

Seems like another good time to bring up Telling More than We Can Know, a nice survey of lots of psychology experiments that showed clear instances of people influenced by a controlled cue, and explaining their behavior without any reference to the cue, but rather spontaneously confabulating reasonable-sounding explanations. Even when we're explaining ourselves, we're often making things up on the fly.

PDF link to the paper: http://www.lps.uci.edu/~johnsonk/philpsych/readings/nisbett.pdf

Direct link: http://www.lps.uci.edu/~johnsonk/philpsych/readings/nisbett.pdf

@Pete Carlton: thanks for the paper, it's a great read!

I think this topic should be explored more in depth here on OB. Is the following conclusion correct? Don't attempt introspection, you will be mislead.

The idea that reason, or deliberative thought, is not the CEO but rather the slave to passions, is David Hume. One argument for the relative idleness of reason was this:

"When we anticipate pain or pleasure from some source, we feel aversion or propensity to that object and “are carry'd to avoid or embrace what will give us” the pain or pleasure (T2.3.3.3).

Our emotion makes us seek the causes of these sources of pain or pleasure, and we use causal reasoning to discover them.

Once we do, our emotion naturally extends itself to those causes, and we act to avoid or embrace them. Plainly the impulse to act does not arise from the reasoning but is only directed by it. "

(from the Stanford Encyclopedia of Philosophy)

http://plato.stanford.edu/entries/hume-moral/#inmo

Let me consult SAI once again (edited conversion)

Me: What is consciousness?

SAI_2100: How many times have I told you Marc! It’s all very simple. Consciousness is analogy formation, equivalent to ontology merging, equivalent to the interface for internal communication (translation) between different types of high level representations (ontological concepts) in the brain. Why on Earth are humans putting out so much noise on something so obvious and really very simple to implement?

Me: So Hanson is on the right track, consciousness is more in the PR (communication) business, than the CEO business?

SAI_2100: Not really. Hanson fails to grasp that while an agents *current* behavior may be largely explained by unconscious processes, current conscious thoughts can influence *future* behavior, via feed-back to the underlying emotional substrate… or , put poetically, ‘feelings follow thoughts’ (nice alliteration ). Causation is bidirectional.

Me: So what is the role of consciousness?

SAI_2100: It’s threefold; (1) High-level filtering, the filtering of irrelevant information via the redirecting of attention, (2) High-level control; high-level modeling of an agents internal states and motivations, , and (3) High-level environment maps; high-level modeling of external environment.

These high-level information summaries cannot be considered as merely ‘just-so’ stories about current brain operations, since this ignores the bi-directional causation I mentioned, the feed-back between high and lower levels of brain operations.

Me: Thanks again SAI

SAI_2100: No problemo, humano!

Ironically, this is pretty much the conclusion my own introspection leads me to: that the running commentary in my head almost always reflects, rather than directs, my decision making processes, and that I have little conscious access to the lower level mechanisms that underlie decisions.

Basically, I ended up with this:

Q: Why do/don't you want to do x?
A: Because I have pleasant/unpleasant feelings when I think about doing it.
Q: Why do you feel that way and not some other way?
A: I have no idea. As far as I can tell, I just do.

Additionally, I've experienced this effect in myself many times.

The comments to this entry are closed.

Less Wrong (sister site)

May 2009

Sun Mon Tue Wed Thu Fri Sat
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30
31