Balancing risk and reward: a rat model of risky decision making

Abstract

We developed a behavioral task in rats to assess the influence of risk of punishment on decision making. Male Long-Evans rats were given choices between pressing a lever to obtain a small, 'safe' food reward and a large food reward associated with risk of punishment (footshock). Each test session consisted of 5 blocks of 10 choice trials, with punishment risk increasing with each consecutive block (0, 25, 50, 75, 100%). Preference for the large, 'risky' reward declined with both increased probability and increased magnitude of punishment, and reward choice was not affected by the level of satiation or the order of risk presentation. Performance in this risky decision-making task was correlated with the degree to which the rats discounted the value of probabilistic rewards, but not delayed rewards. Finally, the acute effects of different doses of amphetamine and cocaine on risky decision making were assessed. Systemic amphetamine administration caused a dose-dependent decrease in choice of the large risky reward (ie, it made rats more risk averse). Cocaine did not cause a shift in reward choice, but instead impaired the rats' sensitivity to changes in punishment risk. These results should prove useful for investigating neuropsychiatric disorders in which risk taking is a prominent feature, such as attention deficit/hyperactivity disorder and addiction.

Figure 1

Effects of shock intensity on reward choice in the risky decision-making task. Groups demonstrated differences in reward preference based on shock intensity, with the higher intensity groups preferring the small, safe reward.

Figure 2

Individual variability in risky decision-making. There was a wide distribution of reward preference in rats in both the 0.35 and 0.4 mA shock intensity groups. Each curve represents data from a single subject.

Figure 3

Effects of satiation on reward choice in the risky decision-making task. Neither one nor 24 hours of free-feeding prior to testing affected reward choice.

Figure 4

Effects of reversal of punishment risks. After the order of risk presentations was reversed, rats continued to demonstrate discounting of the large risky reward in a manner similar to that under ascending risk presentations. The ascending risk data are replottted from Figure 1 (0.35 mA shock intensity) for comparison.

Figure 5

Comparison of the risky decision-making task with other decision-making tasks. a. Performance on the risky decision-making task with 0.35 mA shock intensity. b. Performance on the delay discounting task. c. Performance on the probability discounting task. Insets show scatterplots and regression lines for comparisons of performance on different tasks.

Figure 6

Effects of pharmacological treatments on risky decision-making. a. Rats were tested under the influence of systemic 0.33, 1.0, and 1.5 mg/kg doses of amphetamine. Amphetamine decreased preference for the large risky reward in a dose-dependent fashion, with the 1.5 mg/kg dose differing significantly from saline conditions (p < .05). b. Rats were tested under the influence of systemic 5, 10, and 15 mg/kg cocaine. Rats exposed to cocaine at the 5 and 15 mg/kg doses failed to adjust reward choice as the risk of punishment increased.