Beyond diffusion
Choosing what to do is not just defined by payoffs, social structure and a single social learning strategy, as we modeled it in the previous chapters on diffusion. In this part of the notes we expand the scope of our social behavior theory to include more features: opinions and polarization, uncertainty and its effect on social learning, and a more nuanced cognitive learning model called reinforcement learning.
Opinions (aka beliefs) shape our expectations about the benefits of different behaviors and our openness to learn from others, which in some contexts depends on how similar interaction partners are to one another.
**Environmental uncertainty can reduce, remove, or even reverse the benefits of social learning if environmental variability is unpredictable. If the environment is unpredictable, socially-learned information could become outdated before it is learned, meaning a previously adaptive or sustainable behavior could have become maladptive or unsustainable with the passing of time. For example, monocrop agriculture that tills the ground every season may be less economical and sustainable than it once was given that climate catastrophes occur more frequently, but with unpredictable timing.
Models that flexibly integrate information to update behavioral outcome expectations that guide behavior are called reinforcement learning models. We humans have an extensive repertoire of genetic cognitive adaptations for dealing with this sort of uncertainty. We update our expectations behavioral benefits based on the stream of information we experience in the form of learning from others while accounting for discrepencies between their personal contexts and ours (Witt et al. 2024). Humans are no slaves to socially-learned information, of course, so these models get us one step closer to more realistic social learning.