Abstract
In this chapter, theories of learning are not discussed at all. Numerous texts
exist where they can be found. It would be enough to note that behavior has two
aspects: 1) explorative and 2) exploitative in active inference. The former is sensitive
to ambiguity, and the latter is sensitive to risk. In the absence of ambiguity, active
inference reduces to a Bellman scheme [1]. Bayesian inference is integrated with active
inference in free-energy formulation. Actions are guided by predictions and are refined
by sensory feedback. The variational free energy is a function of observations and a
probability density over their hidden causative agents. The time average of energy is
action. Minimum variational free energy corresponds to a principle of least action.
Perception can be regarded as a minimum of free energy with respect to inbound
sensory information and action as a minimization of free energy with respect to
outbound action.
Synaptic modification is a prerequisite for learning to occur. What one learns must be
preserved for future use. Therefore, it needs to be stored. That storage is memory.
Neural plasticity is the basis for memory formation. Information about biologically
important events (Pavlovian conditional fear, Pavlovian conditioned eye-blink) reach
centers in the amygdala and cerebellum through circuitry, which depends on the
modality of stimulus and its complexity. In the present chapter, memory systems are
introduced to the reader, starting from the Baddeley-Hitch model of working memory.
Working memory is also known as short-term memory (STM). Certain information
stored in short-term memory is transferred to a memory system known as Long-term
memory (LTM). The brain makes decisions as to which information is to be
transferred to LTM. The role of brain oscillations in memory formation is also
discussed. 7±2 rule states that STM in humans can store only 5 pieces of information
when it is complex; on the other hand, it can store 9 pieces when information is simple.
A method to characterize the complexity of information is given. Information transport
in the brain is thoroughly discussed. The chapter ends with a discussion on the
discovery of engram cells, which participate in systems consolidation of memory.