-
In contrast to forcing, introducing an affordance process is a more natural way of doing it . The affordance process represents the external environment’s interaction with the diffusion process. The diffusion process thus, encapsulates the internal dynamics of the animal.
-
Metastability:
-
Time scales : there is a lack of theoretical framework that tie different temporal scales and their contribution to the emergence of behavior.
-
Conceptual of model behavior (control theory, self-organizing processes, closed loop, dynamical system): A conceptual model defines the underlying abstraction used to designate what constitutes a behavior. Within a control theoretic framework, there is a target value / function to which the system tunes itself (it has some teleological soundings). The self organization framework , there is a spontaneous emergence of behaviors that are not driven by a control signal (contrary to the control theoretic process). Our framework belongs to the category of self organization.
-
artificial intelligence: Given that artificial agents will also be embedded in an environment either ”real” or ”artificial”, the implementation of AI - environment loop (relation) is crucial. One way to achieve is via affordances that the AI learns through its interaction with the environment. We avoided using terms that are hard to define like intelligence and one that , unintentionally, put anthropomorphism at the center. Intelligence , with its working definition, is already implied in the action of closed loop selection of affordances in a particular environment.
-
Application in foraging, navigation, vocalization, decision making (drift diffusion): In a real world decision making process, the underlying drift rate changes continuously, our affordance - diffusion model allows us to model the decision making process as a drift - diffusion process in which the drift rate is changing ( according to the affordances in the environment). Same goes for the foraging that is often modeled using random search processes (random walk) and our model can extend the understanding of foraging in terms of the environmental constraints (where there is an affordance process interacting with the environment ) . In spatial navigation tasks, the change in the animals’ behavior given the spatial constraints of the environment is not usually studied. Our model allows us to use arbitrary shapes of potential which extends the applicability of the system to many different behaviors.
-
our affordance - diffusion model clarifies to us why methods used to cluster / classify / segment behavior actually work (most famously nowadays the work of Josh Shaevitz and Gordon Berman)
-
Relation to statistical inference (simulated annealing)
-
What is adaptation
-
non-equilibrium thermodynamics (thermodynamics of prediction, causal entropic forces)
-
Discussion about the constraints