In a new study, researchers offer evidence that chimpanzees rationally revise their beliefs in light of new evidence, a hallmark of epistemic rationality.
More than a century ago, naturalist Henry Smith Williams observed that birds nesting in his garden differed in their colour preferences for nest-building materials. His observations challenge the assumption that individuals within the same species respond similarly to the same colour cues, but the stability and consequences of these apparent preferences remain largely unexplored. Here, we tested the colour preferences of laboratory-raised zebra finches (Taeniopygia guttata) with access to pink and orange nest-building string. Birds were tested under conditions lacking relevant environmental cues, social information, or previous experience with the experimental colours. Individuals differed in both their preferred colour and the strength of preferences, which were more strongly repeatable across short-term than long-term intervals. We next asked whether these stable individual preferences influenced how birds respond to information, reanalyzing published social-learning datasets spanning different informational contexts. Stronger initial preferences generally predicted reduced use of socially demonstrated nonpreferred colours during nest building. These findings suggest that stable individual preferences represent an underappreciated source of variation in how animals acquire and use information during decision making.
This study examined the Weibull model for individual learning curves in Pavlovian conditioning from a theoretical perspective. Although the Weibull model effectively captures variability in individual learning, it remains unclear what the parameters in this model represent. First, we identified similarities in the mathematical representations of the Weibull and error-correction models and showed that two of the three parameters in the Weibull model can be interpreted in the same manner as those in the error-correction model. Next, to explore possible psychological interpretations of the remaining parameter, α , and to examine how α is distributed across individual learning curves, we analyzed four fear-conditioning datasets obtained from both humans and animals. Although it was difficult to establish a definitive interpretation of α in this paper, we discuss possible variables that may be associated with α .
In natural environments, the outcomes of actions are often delayed and separated from the actions themselves by intervening events, posing a credit assignment problem for learning. Although both humans and nonhuman animals can associate actions with immediate subsequent outcomes, little is known about learning when action–outcome delays span multiple, unpredictably ordered events. We investigated whether chimpanzees could associate a choice with a reward delivered several trials later, using a touchscreen-based task. Six chimpanzees completed four experimental rounds involving three trial types: an immediately rewarded discrimination, a non-rewarded choice that triggered a delayed reward, and a reward-delivery trial contingent solely on prior choices. Three chimpanzees showed a significant increase in choosing the stimulus associated with delayed reward. Reversal of stimulus roles in a later round confirmed that increased selection of the delayed-reward stimulus was not attributable to fixed stimulus preferences. Although performance remained variable across individuals and sessions, the results suggest that chimpanzees possess learning mechanisms capable of bridging extended action–outcome gaps, likely operating implicitly rather than through explicit understanding of task structure. Our study contributes to comparative perspectives on the evolutionary origins of learning beyond the “here and now.”
Spatial navigation is considered essential for survival and therefore likely to have been selected for across vertebrates; however, similarities and differences in the mechanisms underlying these abilities across species remain unclear. The current experiment was designed to investigate strategies underlying spatial navigation in green frog (Rana clamitans) tadpoles. Tadpoles were trained to navigate a maze either by using external cues to navigate (place learners), or by remembering the specific egocentric actions that led to escape from the maze (response learners). There were no significant differences in acquisition following place or response learning, which suggests some flexibility in their spatial navigation strategies. However, only ten tadpoles reached the learning criterion before metamorphosis, which was fewer than that predicted by chance alone. Of these, six were place learners and four were response learners. Three place learners appeared to retain this information in the short term - pointing to the intriguing possibility that green frog tadpoles may retain this information for a period of several weeks or longer. Future studies should set a stricter learning criterion and focus on how memory is retained following metamorphosis.
