The spatial context of everyday speech to children is remarkably consistent. Words repeatedly occur in the same locations, and these words are learned earlier than those which are more scattered in use. Yet little is known about how spatial contextualization mediates this relationship. Does more constrained spatial context itself lead to better word learning? Or does it simply correlate with other informative cues in the input? Here, we assess how word learning is influenced by different levels of spatial contextualization in naturalistic scenes. We use different teaching methods (other-directed versus self-guided) as a proxy for distinguishing how the need for inductive inference mediates reliance on spatial context. We found that greater spatial contextualization led to better word learning, but only when inductive inference was needed. These findings suggest that learners can leverage spatial context to support word learning in the absence of rich linguistic input.

During sentence comprehension, listeners form expectationsabout likely structures before they have reached the end of asentence. Children are more likely than adults to ignore late-arriving evidence when it contradicts their initial parse. Whilethis difference is often ascribed to developmental changes inexecutive function, this paper investigates whether statisticalproperties of child-directed speech could be responsible forchildren’s failure to revise temporarily ambiguous sentences.We examined well-studied garden-path sentences andcalculated surprisal values derived from adult and child-directed corpora at each word. For adult corpora, surprisal washighest where the sentence structure was disambiguated. Forchild corpora, however, values at the disambiguating regionwere low relative to other words in the sentence. This suggeststhat for children, the disambiguating words may be statisticallyweak cues to ruling out their original parse, and that inprinciple, the statistics of child-directed speech couldcontribute to children’s difficulty with garden-path sentences.

During conversations, communication partners rapidly assess shared knowledge based on information in utterances. However, little is known about how this process unfolds, particularly when background information is limited such as when talking to strangers. Do spoken utterances provide valid cues to speaker knowledge? To test this, we applied a cultural consensus framework (e.g., Romney et al., 1986), and asked humans vs. large language models (LLMs) to assess speaker similarity based on their transcribed utterances. On each trial, participants saw two language samples that varied in speaker expertise (e.g., A: expert, B: novice) and were asked which one was more similar to a third sample, which was produced by either an expert or novice (X). Accuracy was highest for GPT-4 followed by humans and GPT-3.5. Humans and GPT-4 were more accurate at categorizing language samples from experts, while GPT-3.5 was better with novices. Likewise, humans and GPT-4 were more accurate with samples from adult compared to child speakers, while GPT-3.5 was similar across the two. Item-level performance by humans and GPT-4 was strongly associated, while both were unrelated to GPT-3.5. Our findings suggest that language-based cultural consensus may enable reliable inferences of common ground during communication, providing an algorithmic-level description of how partners may infer states of the world.

Development of cortical regions with precise, sharp, and regular boundaries is essential for physiological function. However, little is known of the mechanisms ensuring these features. Here, we show that determination of the boundary between neocortex and medial entorhinal cortex (MEC), two abutting cortical regions generated from the same progenitor lineage, relies on COUP-TFI (chicken ovalbumin upstream promoter-transcription factor I), a patterning transcription factor with graded expression in cortical progenitors. In contrast with the classical paradigm, we found that increased COUP-TFI expression expands MEC, creating protrusions and disconnected ectopic tissue. We further developed a mathematical model that predicts that neuronal specification and differential cell affinity contribute to the emergence of an instability region and boundary sharpness. Correspondingly, we demonstrated that high expression of COUP-TFI induces MEC cell fate and protocadherin 19 expression. Thus, we conclude that a sharp boundary requires a subtle interplay between patterning transcription factors and differential cell affinity.