Human Judgment [a computational view of decision making]
Human judgment begins with the information bottleneck. The world contains more dimensions and complexity than any individual can encode. The truth as we know it, the things we see, the things we observe, are very high dimensional. We live in a high dimensional world. The human need for simplicity in this complexity, that there may exist simpler rules that we can follow, is an admission of constraints in our memory system, in our intelligence. Not to say that any other machine or tool can exist which will not have such constraints. Yet, what matters here is that there are constraints— human constraints. To manage this overload, observers filter. Choose a subset of data to focus on which maximizes, or optimizes something that they choose. It could be predictability of the future, could be survivability, could be success, could be any variable of their choosing, any optimization function that they want to maximize is up to them. What they end up with is—
a low dimension projection of a high dimensional observed truth or reality.
The compressed data is then processed along and further reduced in dimensionality into a couple of multi label action spaces because most human work is action oriented. Or it creates knowledge for another step of it to be knowledge work which feeds into an action space at some n-th step. However, the point is that humans are compressing knowledge and observation into action and they need a decision space and a decision rule to connect these two: action spaces, low dimensional and costly, and observed spaces, very high dimensional and comparatively free. Making choices and acting on them is time consuming, and costly in many ways of the word. So, the job of the individual brain or social dynamics processes (in case of group decisions) like discussions, talks, meetings, is to make sure that we end up with actions that maximize something. It need not even be a maximization, just satisfaction of some things we're trying to satisfy, like variables that we care about individually or societally.
Human judgment in this way is simply an exercise in dimensionality reduction. I am not commenting on whether this is useful or whether this could be done better. I'm just trying to show the exact process in a computational frame. How the reader interprets this or makes this useful to themselves is not the focus of this piece of writing; that is up to them, and that is another low dimensional judgment that the reader must make.
humans have a limited “context window”
Because cognitive resources are finite, no projection can capture everything . Perhaps it is not even possible to have a system or an observer which is “unbiased”. So far, this has not stopped us. On another note, what makes “unbiased” worth chasing?
