Representational exchange in human social learning: Balancing efficiency and flexibility

Abstract

What makes human social learning so powerful? While past accounts have sometimes prioritized finding the single capacity that makes the largest difference, our social learning abilities span a wide spectrum of capacities from the high-fidelity imitation of behaviors to inferring and learning from hidden mental states. Here, we propose that the power of human social learning lies not within a single capacity, but in our ability to flexibly arbitrate between different computations and to integrate their outputs. In particular, learners can directly copy the demonstrator’s actions in the absence of causal insight (policy imitation), infer their instrumental values (value inference), or infer their model of the world and intrinsic rewards (belief inference and reward inference). Each of these strategies trades off the cost of computation against the flexibility and compositionality of its outputs. Crucially, we have the capacity to arbitrate and exchange information between these representational formats. Human social learning, we suggest, is powerful not just because of the way it moves information between minds, but also because of the way it flexibly moves information within them.

Mentalizing regions and anterior cingulate cortex represent learners’ beliefs during teaching

“…Specifically, teachers chose examples that would be most informative to a learner who interprets them literally-that is, one who rules out the options that are contradicted by the examples provided and is indifferent between the rest. Although coordinated use of recursive mentalizing would have accelerated learning slightly in our task (e.g., in the toy example in Figure 2), it is plausible that teachers may have eschewed this approach due to its computational demands (Hawkins et al, 2021;Wu et al, 2021), or to ensure that their examples are understood by a broad audience of learners of varying degrees of sophistication (Frank & Liu, 2018).…”

Tradition and invention: The bifocal stance theory of cultural evolution

“…Much like a thrifty shopper or an efficient long-distance runner, adaptive organisms should not only maximize rewards, but also account for the cognitive costs of different strategies. While resource-rational adaptations have been widely studied in the context of individual decision making (Kool, Gershman, & Cushman, 2018;Shenhav et al, 2017), we propose that a similar trade-off exists in social learning (Wu, Vélez, & Cushman, 2022).…”

Model-Based Assimilation Transmits and Recombines World Models

“…The representational exchange framework of social learning (Wu, V élez, & Cushman, 2022) describes how different social learning mechanisms form a hierarchy, trading off computational costs against flexibility and compositionality (Fig. 1).…”

“…Whereas action copying operates over motor programs, value inference estimates representations of instrumental value (i.e., goals), requiring an extra layer of inference and incurring putatively greater computational costs (Wu et al, 2022). However, it may offer better generalization to new situations.…”