Oct 21, 2010

environmental sensing

the distributed cognition reading group forges on this week, into donald hoffman's visual intelligence, which stilgoe assigned years ago. now slightly dated, it presents a shortish list of rubrics the mind follows for interpreting optical information acquired from the world without. the point of the book is to convince the reader that vision is both more cognitively involved than it seems and yet also reducible to simpler rules than might at first appear to be the case. it succeeds on both counts.

i liked the book when I first read it mostly because it presented a vast quantity of research in an almost patronisingly accessible fashion. the 34 (i think i'm close) rules of visual image processing were also stunning because i hadn't thought about this stuff prior to reading about it in hoffman and seeing his accompanying illustrations, each designed to highlight the particular rule being discussed. revisiting the book, my original impression of the care taken in writing it remains unchanged. it's science writing of rare quality. the power of the illustrations and the systematic presentation of the rules of visual image processing is diminished somewhat by greater familiarity with the subject matter, but not by much.

my quibbles this time around are with underlying assumptions. most of these are not, in the grand scheme, very important. the one which is worth mentioning is the question of where these rules of visual intelligence come from to start with. hoffman goes on a brief chomsky lovefest some way into the book, and argues that there is something innate about these rules. the argument for innateness (or hardwiring, or whatever) is that the rules are non-obvious, they emerge at a young age, and are almost universal. this argument seems incomplete. perhaps this is because i have, lets face it, no training in the cognitive sciences. but for counterargument's sake, here's another explanation.

a quick survey of the rules leads to the overwhelming conclusion that they are designed to be as parsimonious as possible in terms of cognitive processing. the constraint imposed by the laws of physics produces patterning and regularity in the behaviour of physical objects in the environment as perceived visually. we are pattern-seeking and pattern-identifying beings, and we find patterns in things especially quickly when the patterns are frequently repeated. the ubiquity of the rules of visual information processing therefore plausibly reflects the ubiquity of physical laws rather than innate hardwiring.

(this is not to say that thousands of years of evolution have not produced neural pathways optimized to process these particular types of regularities and patterns, but only to suggest that there may be an antecedent cause for this kind of hardwiring.)

hoffman's conclusion went unappreciated by me when i first read it but not this time around. he notes that the world we perceive is both the world we have interpreted through vision (the phenomenal world) and the world as it "really is" (the relational world). as long as the relational world is systematic in its behaviour, it is possible to interpret what we perceive of it. importantly, the relational world can be completely arbitrary, as long as it is systematic. this is deep stuff, at least to me, for what it contributes to sociological understanding: it goes some way to providing a theoretical basis for understanding why practices that seem totally alien to me can be accepted as norms by someone else in a society where those practices are normal.

clay shirky gave a talk to a small number of people at HBS a few weeks ago about his new book. during the talk, he made much about cultural values and norms, but never defined culture. i brought that up in the Q&A and he instantly gave one of the crispest and best definitions of culture i've heard yet: "culture is the set of habits that allow a group of people to cooperate by assumption rather than by negotiation." i would extend and improve on it and say culture is the set of habits that allow a collectivity to coordinate by assumption rather than by negotiation."

the fact is that we are environment-sensing entities. when we see that the environment is chaotic, our instinct is to keep paying as much attention to it as we can, in as fine-grained as manner as we can, until we can detect a pattern. we are also economical (lazy, if you prefer) so that, if that pattern holds and is detectably stable, we then identify the rule producing the pattern, and use it to reduce attention on that phenomenon to an algorithm and devote our freed-up attention capacity to something else. in doing this, we form habits: sets of rules of cause and effect, sequences of things we do because doing them in the past produced a result that we wanted to recur.

when we act in the world, we combine both the phenomenal world (our perception of the world) and the relational world (the "real" world that responds to our actions). we develop habits by understanding the relationship between action and the behaviour of the relational world in response to action, as mediated through perception. (this can be said, now i think of it, even for how we develop habits for things like scratching a scab. if you haven't tried this, know that it feels really good.) in other words, we develop habits when we identify patterns in how our phenomenal world changes as the relational world responds to the actions we perceive ourselves taking in the relational world.

under all this is the assumption that humans are experimental beings: much of how we act comes from learning how to act in a real world that we perceive as responding to our actions. this is what frustrates me about our continuing obsession with strategy (formal and rational rules) where the really interesting stuff seems to be happening in tactics (practical reason responding to an environment with the resources at hand).

where is the research program on practical reason?

No comments: