Do Children (and Adults) Benefit From a Prediction Error Boost in One-Shot Word Learning?

Influential theories and computational models suggest error-based learning plays an important role in language acquisition: Children learn new words by generating predictions about upcoming utterances and revising those predictions when they are erroneous. Critically, revising stronger (rather than weaker) predictions should further enhance learning. Although previously demonstrated in adults, such prediction error boost has not been conclusively shown in children. To close this gap, we tested 107 participants between the ages of 5 and 10. We found little evidence that word learning in this age group benefits from a prediction error boost. Moreover, we also failed to replicate previous evidence for such an effect in adults. Based on a detailed task analysis, we suggest the variation in adult findings may be partly explained by differences in encoding strategies and that, relatedly, the protracted development of the episodic memory system might explain why children do not experience robust benefits from having stronger (rather than weaker) predictions disconfirmed.

Knowledge of Statistics or Statistical Learning? Readers Prioritize the Statistics of their Native Language Over the Learning of Local Regularities.

A large body of evidence suggests that people spontaneously and implicitly learn about regularities present in the visual input. Although theorized as critical for reading, this ability has been demonstrated mostly with pseudo-fonts or highly atypical artificial words. We tested whether local statistical regularities are extracted from materials that more closely resemble one's native language. In two experiments, Italian speakers saw a set of letter strings modelled on the Italian lexicon and guessed which of these strings were words in a fictitious language and which were foils. Unknown to participants, words could be distinguished from foils based on their average bigram frequency. Surprisingly, in both experiments, we found no evidence that participants relied on this regularity. Instead, lexical decisions were guided by minimal bigram frequency, a cue rooted in participants' native language. We discuss the implications of these findings for accounts of statistical learning and visual word processing.

Lexical access speed and the development of phonological recoding during immediate serial recall.

A recent Registered Replication Report (RRR) of the development of verbal rehearsal during serial recall (Elliott et al., 2021) revealed that children verbalized at younger ages than previously thought (Flavell et al., 1966), but did not identify sources of individual differences. Here we use mediation analysis to reanalyze data from the 934 children ranging from 5 to 10 years old from the RRR for that purpose. From ages 5 to 7, the time taken for a child to label pictures (i.e. isolated naming speed) predicted the child's spontaneous use of labels during a visually-presented serial reconstruction task, despite no need for spoken responses. For 6- and 7-year-olds, isolated naming speed also predicted recall. The degree to which verbalization mediated the relation between isolated naming speed and recall changed across development. All relations dissipated by age 10. The same general pattern was observed in an exploratory analysis of delayed recall for which greater demands are placed on rehearsal for item maintenance. Overall, our findings suggest that spontaneous phonological recoding during a standard short-term memory task emerges around age 5, increases in efficiency during the early elementary school years, and is sufficiently automatic by age 10 to support immediate serial recall in most children. Moreover, the findings highlight the need to distinguish between phonological recoding and rehearsal in developmental studies of short-term memory.

Does it pay to imitate? No evidence for social gains from lexical imitation.

According to an influential hypothesis, people imitate motor movements to foster social interactions. Could imitation of language serve a similar function? We investigated this question in two pre-registered experiments. In Experiment 1, participants were asked to alternate naming pictures and matching pictures to a name provided by a partner. Crucially, and unknown to participants, the partner was in fact a computer program which in one group produced the same names as previously used by the participant, and in the other group consistently produced different names. We found no difference in how the two groups evaluated the partner or the interaction and no difference in their willingness to cooperate with the partner. In Experiment 2, we made the task more similar to natural interactions by adding a stage in which a participant and the partner introduced themselves to each other and included a measure of the participant's autistic traits. Once again, we found no effects of being imitated. We discuss how these null results may inform imitation research.

The role of language production in making predictions during comprehension.

Language comprehension depends heavily upon prediction, but how predictions are generated remains poorly understood. Several recent theories propose that these predictions are in fact generated by the language production system. Here, we directly test this claim. Participants read sentence contexts that either were or were not highly predictive of a final word, and we measured how quickly participants recognised that final word (Experiment 1), named that final word (Experiment 2), or used that word to name a picture (Experiment 3). We manipulated engagement of the production system by asking participants to read the sentence contexts either aloud or silently. Across the experiments, participants responded more quickly following highly predictive contexts. Importantly, the effect of contextual predictability was greater when participants had read the sentence contexts aloud rather than silently, a finding that was significant in Experiment 3, marginally significant in Experiment 2, and again significant in combined analyses of Experiments 1-3. These results indicate that language production (as used in reading aloud) can be used to facilitate prediction. We consider whether prediction benefits from production only in particular contexts and discuss the theoretical implications of our evidence.

Making oneself predictable in linguistic interactions.

While language production is a highly demanding task, conversational partners are known to coordinate their turns with striking precision. Among the mechanisms that allow them to do so is listeners' ability to predict what the speaker will say, and thus to prepare their response in advance. But do speakers also play a role in facilitating coordination? We hypothesized that speakers contribute by using coordination smoothers - in particular by making their turns easier to predict. To test this, we asked participants to type definitions for common English words, either on their own (n = 26 individuals) or interacting with a partner (n = 18 pairs), and we measured the timing with which they produced the definitions. In a post-test, additional participants (n = 55) attempted to predict the final word of these definitions and rated them for quality. We found that interacting speakers initiated their turns with less variable delays than solo individuals. In contrast, our post-test measures suggested that jointly produced definitions were in fact of lower predictability and quality than those produced by individuals, but the analysis revealed these findings were likely confounded by task difficulty. We propose that the reduction in temporal variability observed for interacting speakers may facilitate prediction and thus act as a coordination smoother in linguistic interactions.

Action-effect anticipation and temporal adaptation in social interactions.

Interacting agents may anticipate their partner's upcoming response and include it in their action plan. In turn, observing an overt response can trigger agents to adapt. But although anticipation and adaptation are known to shape action control, their interplay in social interactions remains largely unexplored. In 4 experiments, we asked how both of these mechanisms could contribute to one striking phenomenon: Agents initiate actions faster when they know their partner will produce a compatible rather than an incompatible response. In Experiment 1, we manipulated the compatibility between agents' actions and partners' responses and investigated the interplay between adaptation and anticipation within the same dyadic interaction. In Experiments 2-4, we isolated the contribution of each mechanism by having agents interact with virtual partners whose responses could be experimentally controlled. We found that adaptation and anticipation exert parallel but independent effects on action execution: Participants initiated their actions more quickly when the upcoming partner response was compatible and, independently, when their partner had responded more quickly on the preceding trial. These findings elucidate models of action control in social interactions. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

Gaze-based rehearsal in children under 7: a developmental investigation of eye movements during a serial spatial memory task.

The emergence of strategic verbal rehearsal at around 7 years of age is widely considered a major milestone in descriptions of the development of short-term memory across childhood. Likewise, rehearsal is believed by many to be a crucial factor in explaining why memory improves with age. This apparent qualitative shift in mnemonic processes has also been characterized as a shift from passive visual to more active verbal mnemonic strategy use, but no investigation of the development of overt spatial rehearsal has informed this explanation. We measured serial spatial order reconstruction in adults and groups of children 5-7 years old and 8-11 years old, while recording their eye movements. Children, particularly the youngest children, overtly fixated late-list spatial positions longer than adults, suggesting that younger children are less likely to engage in covert rehearsal during stimulus presentation than older children and adults. However, during retention the youngest children overtly fixated more of the to-be-remembered sequences than any other group, which is inconsistent with the idea that children do nothing to try to remember. Altogether, these data are inconsistent with the notion that children under 7 do not engage in any attempts to remember. They are most consistent with proposals that children's style of remembering shifts around age 7 from reactive cue-driven methods to proactive, covert methods, which may include cumulative rehearsal.

Spontaneous adaptation explains why people act faster when being imitated.

The human ability to perform joint actions is often attributed to high-level cognitive processes. For example, the finding that action leaders act faster when imitated by their partners has been interpreted as evidence for anticipation of the other's actions (Pfister, Dignath, Hommel, & Kunde, 2013). In two experiments, we showed that a low-level mechanism can account for this finding. Action leaders were faster when imitated than when counterimitated, but only if they could observe their partner's actions (Exp. 1). Crucially, when due to our manipulation the partner's imitative actions became slower than the counterimitative actions, leaders also became slower when they were imitated, and faster when counterimitated (Exp. 2). Our results suggest that spontaneous temporal adaptation is a key mechanism in joint action tasks. We argue for a reconsideration of other phenomena that have traditionally been attributed solely to high-level processes.