Attention in Category Learning
Appreciating that varied stimuli belong to different categories requires attention to be differentially allocated to relevant and irrelevant features of those stimuli. We have introduced new experimental (“peck tracking”) and computational tools for assessing attention in pigeons both during and after category learning.
During Learning
In several of our experiments, when a category exemplar (see two examples of Categories A and B on the left) was presented on the computer screen, pigeons had to peck it multiple times in order to move forward with the task. The pigeons were free to peck any of the features, relevant or irrelevant. We recorded the location of the pigeons’ pecks in order to determine whether or not they selectively directed their pecks to the relevant features of the category exemplars.
As training proceeded, categorization accuracy increased, as expected. Moreover, the pigeons also increasingly pecked the relevant stimulus features, which suggests that they were tracking the relevant information for solving the task.
Also, the pigeons’ accuracy was considerably higher after they pecked the relevant category features than after they pecked the irrelevant features. Only tracking the relevant information resulted in successful classification of the category exemplars.
Wasserman, E. A., & Castro, L. (2021). Assessing attention in category learning by animals. Current Directions in Psychological Science, 30, 495-502.
After Learning
Other experiments revealed that, compared with human adults, pigeons do not selectively attend to perfectly relevant (or deterministic) features only. It is important to note that a category discrimination can be accomplished by perceiving the family resemblance of the exemplars in each category. In this case, attention may be widely distributed among multiple features. We will call these features probabilistic, because none of them can predict the correct category all the time, but they tend to be more predictive of one category than another.
We gave human adults and pigeons a categorization task that could be learned on the basis of either one deterministic feature (showing selective attention) or multiple probabilistic features (showing distributed attention). Although both humans and pigeons successfully solved the task, computational modeling after task mastery revealed that human adults focused their attention on deterministic information and filtered less predictive information; pigeons, on the other hand, tended to distribute their attention among several features.
Castro, L., Savic, O., Navarro, V., Sloutsky, V. M., & Wasserman, E. A. (2020). Selective and distributed attention in human and pigeon category learning. Cognition, 204, 104350. https://doi.org/10.1016/j.cognition.2020.104350
Rule-based and Associative Strategies
If pigeons are such proficient classifiers of complex visual stimuli, then how do they do it? Do they use elaborate rules as people do? Or are they powerful computing machines deploying basic associative processes?
Some categorization tasks may be solved by deploying declarative rules (rule-based, RB tasks) or by creating straight associations between specific stimuli and the correct response to them. Other tasks require integrating information (information integration, II tasks) from different dimensions, so that generating a verbal rule becomes quite difficult. In that case, association of the stimuli to their responses may be a more efficient way to learn them.
We have found that pigeons solve both RB and II tasks using associative mechanisms, whereas humans solve RB tasks using declarative rules. Also, humans tend to deploy declarative mechanisms for II tasks in the early stages of learning, even when this delays their learning.
In addition, despite showing no evidence of declarative processing, pigeons robustly succeeded at tasks that challenged even human learners, suggesting that associative learning mechanisms merit far greater attention than they often receive.
O’Donoghue, E. M., Broschard, M. B., Freeman, J. F., & Wasserman, E. A. (2022). The Lords of the Rings: People and pigeons take different paths mastering the Concentric-Rings categorization task. Cognition, 218, 104920.