Building Reward Models

Building Reward Models: Mathematical Models for Reward

Reward models of Rescorla & Wagner (1972) and Sutton & Barto’s (1990; 1998) have made appreciable contributions to psychology and computer science. Rescorla & Wagner’s (1972) reward learning model used rate parameters, which monitored stimulus salience and the breadth of learning characterizing both the unconditioned stimulus (US) and conditioned stimulus (CS). Their equation reflected the US’s ability for influencing CS stimulus qualities. They termed this imputed rewarding influence as gains to the CS’s associative strength. (Read more)

:

Appraisal Models-Scherer & Colleagues’ SEC of Emotion

Scherer and colleages developed a primary and secondary appraisal model later called stimulus evaluation checks (SEC) of emotion.  These checks were “presumed to underlie the emotion-constituent appraisal process”.  SECs were initially defined as monitoring mechanisms that inspected, appraised, and evaluated the environment for perceptual and environmental input important for eventually yielding expected outcome.  These processes eventually gave “rise to different emotions” (Read more)

:

Appraisal Models of Others

An appraisal is expressed on either side of an emotion. It has causative components that first elicits the expression of emotion and descriptive components describing an emotional state by its core relational theme (CRT).  According to Parkinson’s conceptualization, statements like “I feel loss”  and “I feel helpless” serve to measure an earlier expression of emotion (like sadness).  This is in contrast to a primary appraisal as being a reactive conceptualization of the impact of environmental outcome on the self and a precursor of the expression of an emotion, like sadness. (Read more)

:

Current Emotional Theories & Models

Ahn & Picard recognized the significant roles that emotions, intrinsic and extrinsic rewards,  and motivations played in learning and in decision-making. They identified extrinsic rewards as stimulus salience and intrinsic rewards to the “wanting circuitry.”

Their affect model was also influenced by the valuation of reward, R0 , and the outcome response of not receiving it,  i.e. the state pf feeling bad with breached expectation, and the state of receiving it, i.e. the feeling good state. Affective probabilities were represented. 

Malfaz & Salichs developed a computer model for an autonomous robot (AI). (Read more)

: