I know, I know…I promised to write a second part to this post like a month ago, but it’s been a very busy summer. Aside from work in two labs, the Mrs. is preggers, and we’re running around like crazy trying to get the house ready and so forth. While spending as much free time as possible ruminating over how to fully realize my discipline-uniting treatise, I’ve had another idea that I feel merits discussion:
In thinking about the unity of anthropological inquiry, I’ve taken the view of the individual as a product of and node in many overlapping networks (biological, neurological, cultural, linguistic, historical). The behavior and ideas of any one individual, then, is the product of the action of those networks at their specific node. Thus, the almost mystic force of the combined networks compels behavior, and behavior is essentially output back to these networks.
This view of human behavior is rather close to what I’d imagine many sciencey-types believe: that human behavior, while infinitely complex, is largely the result of biology and culture, instinct and learning, nature AND nurture. Key to this kind of view–or any view for that matter–is the idea that past experiences inform the decisions we make; we learn or are conditioned during events in our lives to change our behavior, and somehow call on that experience to modulate our response in a given situation, generally to increasingly optimal effect. The assumption is that our memory of events long-past informs the decisions we make today.
But what if this weren’t the case? What if our behavior was essentially only governed by the immediate state of our body and mind? What if a person’s behavior was not a result of experientially-modified neurological action and output, but of the state of the system (i.e. the individual within their immediate environment) at that particular moment? We’ve all had an experience where we say or do something and then think “I know better than to do/say that.” You “know better,” but you react in a manner that surprises even you.
A Hidden Markov Model (HMM), in my very general knowledge thereof, is essentially a system where the process by which output is generated is “hidden” to the observer, and furthermore, each step in the progression that results in any particular output is dependent only upon the previous step, or the “state” of the system at that point, but the state transitions are unknown to the observer. I’ll use an analogy I “modified” from Wikipedia: Imagine a wall with a screen that displays a long series of numbers, and behind the wall, there is a genie who draws a numbered ball from each of three urns with a known mix of balls* and types each number in for display on the screen. The genie has a process to determine the order in which he will draw from the urns, which involves plugging the number of the previous pot into some kind of equation with another number drawn at random. Thus, the selection of the pot for nth number is determined by the pot for the (n– 1)th number; the selection does NOT directly depend upon the previous selections, only the last (see Fig. 1) That is a Markov Model, and it is hidden in that anyone who sees the screen cannot see the process of number selection.
The process states here involve moment-to-moment variations in neurological and endocrine activity, and the immediate environment experienced by the individual (the setting, the people, etc.) might be analogous to the equation/random number that the genie uses to choose urns**. If behavior is ultimately the result of a Markov process, the complexity and unpredictability of human behavior would be a result of the sheer number of potential process states (urns) that could result in a particular output. One might ask if we know the sets of numbered balls in the urns, i.e., the potential outputs? Are potential outputs limited? Does language limit potential outputs, for example, in keeping with Sapir (1933)and Whorf (1939)?
Of course, no analogy is apt. Human beings obviously plan for the future and execute those plans. People have memories, and we likely have them for a reason. However, I am not arguing that memory and learning aren’t important, I’m proposing that memories, etc. might not have an influence on human behavior minute-to-minute.
* What would Edward Said think?!
** Because the visible environment plays a role, the model or process that results in behavioral output is not entirely hidden.