Transmission Versus Truth, Imitation Versus Innovation: What Children Can Do That Large Language and Language-and-Vision Models Cannot (Yet).

Much discussion about large language models and language-and-vision models has focused on whether these models are intelligent agents. We present an alternative perspective. First, we argue that these artificial intelligence (AI) models are cultural technologies that enhance cultural transmission and are efficient and powerful imitation engines. Second, we explore what AI models can tell us about imitation and innovation by testing whether they can be used to discover new tools and novel causal structures and contrasting their responses with those of human children. Our work serves as a first step in determining which particular representations and competences, as well as which kinds of knowledge or skill, can be derived from particular learning techniques and data. In particular, we explore which kinds of cognitive capacities can be enabled by statistical analysis of large-scale linguistic data. Critically, our findings suggest that machines may need more than large-scale language and image data to allow the kinds of innovation that a small child can produce.

Effects of social context manipulation on dorsal and ventral hippocampal neuronal responses.

The hippocampus is critical for contextual memory and has recently been implicated in various kinds of social memory. Traditionally, studies of hippocampal context coding have manipulated elements of the background environment, such as the shape and color of the apparatus. These manipulations produce large shifts in the spatial firing patterns, a phenomenon known as remapping. These findings suggest that the hippocampus encodes and differentiates contexts by generating unique spatial firing patterns for each environment a subject encounters. However, we do not know whether the hippocampus encodes social contexts defined by the presence of particular conspecifics. We examined this by exposing rats to a series of manipulations of the social context, including the presence of familiar male, unfamiliar male and female conspecifics, in order to determine whether remapping is a plausible mechanism for encoding socially-defined contexts. Because the dorsal and ventral regions of the hippocampus are thought to play different roles in spatial and social cognition, we recorded neurons in both regions. Surprisingly, we found little evidence of remapping in response to manipulation of the social context in either the dorsal or ventral hippocampus, although we saw typical remapping in response to changing the background color. This result suggests that remapping is not the primary mechanism for encoding different social contexts. However, we found that a subset of hippocampal neurons fired selectively near the cages that contained the conspecifics, and these responses were most prevalent in the ventral hippocampus. We also found a striking increase in the spatial information content of ventral hippocampal firing patterns. These results indicate that the ventral hippocampus is sensitive to changes in the social context and neurons that respond selectively near the conspecific cages could play an important, if not fully understood role in encoding the conjunction of conspecifics, their location and the environment.