A review of important findings from delay discounting research (Article I). Effects of increasing, decreasing, and constant delays of reinforcement on choice behavior in rats (Article II)


Publication date


Series/Report no



Høgskolen i Oslo og Akershus

Document type


Master i læring i komplekse systemer


The subject of delay discounting has been widely studied within the frames of behavioral science. Human and non-humans have been shown to increasingly prefer a smaller sooner reward over a larger later one, when the delay preceding the last is increased. A hyperbolic function has proven effective in describing delay discounting in both humans and non-humans. Further, a number of factors have been found to affect rates of discounting. For example, children tend to discount delayed rewards at a higher rate than adults and small rewards are discounted more steeply than large rewards. Delay discounting has also been investigated as a measure of impulsivity: choosing a smaller sooner reward over a larger later reward can be referred to as impulsive choice behavior. Another area that has received considerable attention is the connection between addiction and higher discounting rates of delayed outcomes. People suffering from various addictions have been found to show a greater preference for immediate rewards over larger delayed rewards that control participants without reported addictions. Article I reviews some of the published literature on delay discounting and discusses important findings in delay discounting research. Article II consists of an empirical study that investigates the effects of gradually increasing or decreasing the delay preceding a larger reward as well as keeping the delay constant. This study was conducted with four Wistar rats. Results show that the percentage of responses leading to the larger delayed reward decreased as the delay preceding the delivery of this reward increased. However, this response percentage did not increase as the delay decreased, nor stabilize as the delay was kept constant. These results indicate that responding might have been affected by other variables (such as previous conditions) than the current contingen


Permanent URL (for citation purposes)

  • http://hdl.handle.net/10642/1435