Why don’t guiding cues always guide in behavior chains?

Effects of immediate-reinforcement training on delay discounting behavior in rats

Chronic exposure to delayed reinforcement has been shown to increase choice for larger, later reinforcement in a subsequent delay discounting task. In the 3 experiments presented in this paper, the opposite was tested: effects of chronic exposure to immediate reinforcement on choice in a subsequent delay discounting task. In Experiment 1, larger, later reinforcement choice was significantly reduced as a result of exposure to immediate reinforcement, compared to a maturation/handle control group, in experienced, male Lewis rats. In Experiment 2, with naive male and female Wistar rats, and Experiment 3, with naïve male Long Evans rats, the impact of exposure to immediate reinforcement was less robust, but directionally consistent with results from Experiment 1. These results align with some previous work reporting that exposure to immediate reinforcement may decrease choice for larger, later reinforcement in a delay discounting task, and/or blunt maturational increases in choice for larger, later reinforcement. These findings have implications for future research investigating experience-based interventions to manipulate delay discounting behavior. They also have clinical implications for understanding and treating disorders involving maladaptive choice.

Reduced Frequency of Knowledge of Results Enhances Acquisition of Skills in Rats as in Humans

Macphail's (1985) null hypothesis challenged researchers to demonstrate any differences in intelligence between vertebrate species. Rather than focus on differences, we asked whether rats would show the same unexpected, counterintuitive features of skill learning observed in humans: Factors that degrade performance during acquisition often enhance performance in a subsequent retention/autonomy phase. Providing post-trial "knowledge of results" (KR) on 30-67% of trials instead of 100% degrades accuracy, yet increases retention in a subsequent phase without KR. We tested this feature by providing three groups of rats with KR on every trial (100% KR), 67% KR, or 0% KR. We also provided operant feedback in every trial for completing the left-right lever-press skill (food for correct sequences, timeout for all others). In the autonomy phase, we assessed their ability to complete the skill independently-in the absence of differential cues and KR feedback. In agreement with human performance in the autonomy phase, 67% KR yielded higher skill accuracy than providing 100% KR. Also, providing 67% KR improved skill accuracy above that observed with operant feedback alone (0% KR). Rather than degrading performance during acquisition, the 67% KR condition yielded unexpected higher accuracy than the other conditions. Accuracy increased systematically across our extended acquisition phase, which provided each rat with over 3600 trials compared to 20-30 trials for human studies. Providing limited KR promoted skill learning in rats as it does in humans, consistent with the conjecture that both species share common learning processes. Introducing difficulties to rats during training improved their autonomy.