On the neural computation of utility: implications from studies of brain stimulation reward

Abstract

1. Like other vertebrates, from goldfish to humans, rats will work in order to deliver electrical stimulation to certain brain sites. Although the stimulation produces no evident physiological benefit, it is sought out avidly, as if it were a biologically significant resource. Thus, it has long been thought that the rewarding stimulation activates neural circuitry involved in the evaluation and selection of goals. 2. Computing the utility of goal objects involves a tightly integrated set of perceptual, cognitive, and motivational mechanisms. I argue that rewarding electrical brain stimulation engages only a subset of these mechanisms. If so, comparison of the ways in which the utility of electrical brain stimulation and natural reinforcers are computed may highlight operating principles and isolate components of the computational mechanisms. 3. In the view proposed here, information about goal objects and consummatory acts is processed, in parallel, in three different channels. 3.1. Perceptual processing indicates what and where the goal object is. 3.2. A stopwatch-like interval timer predicts when or how often the goal object will be available. 3.3. Under the influence of information about the current physiological state, an evaluative channel returns a subjective weighting of strength variables such as the concentration of a sucrose solution or the temperature of an air current. 3.4. The output of these channels is recorded in multidimensional records that include 3.4.1. information of perceptual origin about amount and kind (e.g., food, water,or salt) 3.4.2. information from the timer about rate and delay 3.4.3. a subjective assessment of intensity provided by the evaluative channel 4. This chapter addresses the relationships between brain stimulation reward (BSR), the perceptual, interval timing, and evaluative channels, and the variants of utility proposed by Kahneman and his coworkers on the basis of their studies of evaluation and choice in human subjects. 4.1. It is argued that the output of the evaluative channel can be manifested in experience as pleasure or suffering but that awareness is not necessary in order for this signal to influence action. 5. The neural signal injected by rewarding electrical stimulation is portrayed as providing meaningful information about rate, delay and intensity but not about amount or kind. This proposal is used to account for 5.1. competition and summation between BSR and natural rewards 5.2. differential effects of physiological feedback on the utility of BSR and natural rewards 5.3. matching of behavioral allocation to the relative rates and intensities of BSR 5.4. differences in the elasticity of demand for BSR and food in a closed economy 5.5. the high substitutability of BSR for food and water in an open economy 6. The powerful aftereffect of BSR that potentiates efforts to obtain additional stimulation is related to expectancy.