Imagine that we had an infinite amount time and unlimited access to any and all sources of information. What parameters or predicates would we need to account for and predict human behavior on both short and long time scales and in real-life, complex situations?
Whatever parameters we select, let’s call the model “The Seldon Model” after the famous fictional character, Hari Seldon from the Issac Asimov Foundation novels.
Recall that Hari Seldon was a genius psychohistorian that developed an algorithmic science that enabled him to predict in probabilistic terms the future changes, and ultimately, the demise of, civilization. Unfornunately, Seldon’s model only worked for large groups of people. Nonetheless, if we pretend that the Seldon Model was real, what supplemental parameters would we include to extend the model to individuals?
Read on to learn more about some parameters already at work in the cognitive science of humans.
Four possible paramters in the Seldon Model
The Drift Diffusion Model (DDM) has been applied in cognitive science to observations of human behavior in two-alternative forced choice tasks. These tasks typically ask participants to select one of two options dependent on some criteria in a speeded fashion. By applying the DDM model, scientists have been able to explain some of their observations of the time it takes to make a decision in these tasks. Importantly, the model’s success is partly explained by its reliance on just four parameters, a) the amount of information that must be gather prior to reaching a decision, b) the rate at which evidence is accumulated over time, c) how biased they are toward one of the two options, and d) the amount of time they spend on non-decision related activities such as sensory transduction. Thus, by taking these four parameters into account, educators, engineers, and businesses can strategize about things like, optimal test times, product design, and or advertisement purchasing.
Despite the benefits of considering the parameters at work in the DDM model, there are reasons why our Seldon Model might ignore them. Consider, for example, that the temporal deadlines for decisions in these laboratory experiments are unnatural, usually confined to the seconds range, rather than minutes, and many human decisions, like buying a car from a choice of 10 cars, or buying stock from a selection of 1000 stock, making a move in a game of chess, etc. happen on a much longer time scale. Thus, we cam return to our original question. Is there a model that can predict the variations among people for decisions that require BOTH short and long temporal ranges?
This is all just speculation, but I think the Seldon Model would need to also include things like sleep quality, food intake, exercise, relationship quality, and many more variables derived from long term human behavior. However, since people differ in these variables, and since it’s really difficult to track all of these datas ources (not to mention store it at scale), it’s unlikely that any single scientist, organization, institution, or business could do more than speculate about which parameters are necessary in the Seldon Model.
What do you think? Are we closer than before in approximating the Seldon Model? How about the engineers of Google and Amazon? Are they making any headway?