04 August 2015

Cognition as a Collaboration with Nature


In the previous posts' parable and examination on trust and collaboration, the group was always two or more intelligent agents.  An interesting angle to think about is to recast those person-vs-person games into a person-vs-nature story, from which we might learn that intelligence cannot figure out the world without assuming by charity that nature must be out to help it (rather than out to harm it).

This post is extremely speculative, but is an interesting thought experiment, one might say.

Consider, if an animal acts in the world only to get feedback in the form of a veto vs no-veto, as in the form of die or not-die, well that at first seems like how evolution works but it ignores a lot of information the animal receives from nature beyond die-or-not.

After all, there's no a priori reason why an animal should think that walking off a ledge should kill it, and yet it learns not to along with lots of other stuff to do or not do.  Not all of that is encoded genetically via evolution: certainly the appearance of every ledge or hilltop or predator can't all be encoded into DNA. Some basics are, of course, but much is learned once born.  How does it learn it if the only feedback it can ever get is die-or-not?


Obviously species and animals get more information from nature than that one bit of data, but hypothetically, what if we look at nature as a communicating agent rather than as a passive thing to be perceived by an animal.  Imagine it as a non-zero-sum cooperative repeated game.  If
nature means flowers, e.g., then the interaction with bees can be seen as a non-zero-sum cooperative repeated game that the two species play together.  Evolutionary biology has used game theory to model that sort of thing before, so that's nothing new.

Beyond using game theory or the like as a model, what if it's in some way actually true to the nature of bees and flowers, especially as bees are intelligent beings to a degree.  The flowers has to communicate to the bees what it prefers in some way so that the bees can do right by the flowers, and vice versa.

A simple veto-or-not against what the bee does, e.g. by killing them off when they don't do things right, could work in that it would eventually encode what the bees needs to know via evolution into the bee's DNA. But as per the previous parable, it's one bit of information at a time, and extremely inefficient as the bees only has access to essentially randomizing the DNA via mutation and sexual reproduction to test out different hypotheses.

If the bees were to figure out flowers (and nature) more efficiently, it could collaborate with the flowers and communicate mutually their preferences.  That way, both parties could work towards a better Pareto outcome with a higher collective payoff.  That would all require communication between the bees and the flowers, and the bees would have to in some sense assume the flowers (and nature) aren't inherently out to hurt them, that the flowers are being charitable and are "trying" to tell them something helpful, or else it would be hard for them to collaborate (as discussed in the previous parable and examination on collaboration).

We'd also have to see flowers (and nature) as though they were agents capable of making choices of course.  If bees, whether individually or as a group, can be seen as having enough control to have some degree of agency, there's no reason to deny the same to flowers, or indeed to nature.  We're not talking about some advanced moral agency here, just enough to satisfy certain foundations for game theory to apply.

Under those assumptions, the bees would have to play a game of interpretation: taking what is perceived from nature and interpreting it into useful information.  Like the flowers location relative to the sun, and somehow synthesizing that into a (e.g. honey bee) dance that other bees can interpret and then act usefully upon.

How does that interpretation work?  There must be a system of taking immediate perceptions of nature, which are then turned into ever higher levels of perceptualizations (probably made up of ever higher levels of perceptual based features as in the ever more successful Deep Artificial Neural Network computational models of late).  Those high level perceptualizations have to somehow be turned into persistent data somehow stored so that they can effect further cognitive changes or external behaviour changes independent of the original perceptions, for the original parts of nature that was perceived may be long gone!

Probably and in broad terms, those high level perceptualizations, which are directly stimulated from parts of nature that was perceived, are turned into conceptualizations, and conceptualizations are paired up with other conceptualizations to form symbols (in the sense used in cognitive grammar).  The whole process of symbolization of what is perceived thus allows in one shot, e.g., the pairing of "blue sky" to "nice day" (for activity).

In learning language, the words "blue sky" as uttered may be perceptualized then conceptualized then paired with the conceptualization of a physical blue sky that was separately perceptualized then conceptualized.  And so forth with "nice day".  So when told the meaning of why the sky is blue, the sky's blueness can hold immediate meaning of it being a nice day: even if no blue sky is in sight to be
perceived.

I.e. it is the conceptualizations that are persisted and that are used to reason with, where "reason" is used loosely (i.e. not necessarily logically).  Notice the reasoning from blue sky to nice day in the above example occurs as a form of answering the question, "what does that mean", rather than forming propositions and following logical deductions.  One may argue and reframe the reasoning in terms of propositional logical deductions (or the like), but that would seem to be putting things backwards: it's obvious people grow up seeing and seeking "meaning", even and mostly without formal education on logical deductive methods, and so it would seem more likely that "meaning" has more to do with cognition and survival than formal logical methods taught in school.

Anyway, the kind of reasoning that happens within the cognition of an agent, or any kind of reasoning really, is only valuable for survival if the animal can interpret some useful information from nature.  So perhaps to understand cognition, it must be understood in context with the nature that it collaborates with to survive.  Just as to understand honey bees and their dance, it must be understood in context with the flowers and nature that it lives with as a mutually benefiting ecosystem.

No comments: