Open Mind最新文献

Transfer learning, the reuse of newly acquired knowledge under novel circumstances, is a critical hallmark of human intelligence that has frequently been pitted against the capacities of artificial learning agents. Yet, the computations relevant to transfer learning have been little investigated in humans. The benefit of efficient inductive biases (meta-level constraints that shape learning, often referred as priors in the Bayesian learning approach), has been both theoretically and experimentally established. Efficiency of inductive biases depends on their capacity to generalize earlier experiences. We argue that successful transfer learning upon task acquisition is ensured by updating inductive biases and transfer of knowledge hinges upon capturing the structure of the task in the inductive bias that can be reused in novel tasks. To explore this, we trained participants on a non-trivial visual stimulus sequence task (Alternating Serial Response Times, ASRT); during the Training phase, participants were exposed to one specific sequence for multiple days, then on the Transfer phase, the sequence changed, while the underlying structure of the task remained the same. Our results show that beyond the acquisition of the stimulus sequence, our participants were also able to update their inductive biases. Acquisition of the new sequence was considerably sped up by earlier exposure but this enhancement was specific to individuals showing signatures of abandoning initial inductive biases. Enhancement of learning was reflected in the development of a new internal model. Additionally, our findings highlight the ability of participants to construct an inventory of internal models and alternate between them based on environmental demands. Further, investigation of the behavior during transfer revealed that it is the subjective internal model of individuals that can predict the transfer across tasks. Our results demonstrate that even imperfect learning in a challenging environment helps learning in a new context by reusing the subjective and partial knowledge about environmental regularities.

Morality is central to social well-being and cognition, and moral lexicon is a key device for human communication of moral concepts and experiences. How was the moral lexicon formed? We explore this open question and hypothesize that words evolved to take on abstract moral meanings from concrete and grounded experiences. We test this hypothesis by analyzing semantic change and formation of over 800 words from the English Moral Foundations Dictionary and the Historical Thesaurus of English over the past hundreds of years. Across historical text corpora and dictionaries, we discover concrete-to-abstract shifts as words acquire moral meaning, in contrast with the broad observation that words become more concrete over time. Furthermore, we find that compound moral words tend to be derived from a concrete-to-abstract shift from their constituents, and this derivational property is more prominent in moral words compared to alternative compound words when word frequency is controlled for. We suggest that evolution of the moral lexicon depends on systematic metaphorical mappings from concrete domains to the moral domain. Our results provide large-scale evidence for the role of metaphor in shaping the historical development of the English moral lexicon.

A large program of research has aimed to ground large-scale cultural phenomena in processes taking place within individual minds. For example, investigating whether individual agents equipped with the right social learning strategies can enable cumulative cultural evolution given long enough time horizons. However, this approach often omits the critical group-level processes that mediate between individual agents and multi-generational societies. Here, we argue that interacting groups are a necessary and explanatory level of analysis, linking individual and collective intelligence through two characteristic feedback loops. In the first loop, more sophisticated individual-level social learning mechanisms based on Theory of Mind facilitate group-level complementarity, allowing distributed knowledge to be compositionally recombined in groups; these group-level innovations, in turn, ease the cognitive load on individuals. In the second loop, societal-level processes of cumulative culture provide groups with new cognitive technologies, including shared language and conceptual abstractions, which set in motion new group-level processes to further coordinate, recombine, and innovate. Taken together, these cycles establish group-level interaction as a dual engine of intelligence, catalyzing both individual cognition and cumulative culture.

Researchers have recently argued that the capabilities of Large Language Models (LLMs) can provide new insights into longstanding debates about the role of learning and/or innateness in the development and evolution of human language. Here, we argue on two grounds that LLMs alone tell us very little about human language and cognition in terms of acquisition and evolution. First, any similarities between human language and the output of LLMs are purely functional. Borrowing the "four questions" framework from ethology, we argue that what LLMs do is superficially similar, but how they do it is not. In contrast to the rich multimodal data humans leverage in interactive language learning, LLMs rely on immersive exposure to vastly greater quantities of unimodal text data, with recent multimodal efforts built upon mappings between images and text. Second, turning to functional similarities between human language and LLM output, we show that human linguistic behavior is much broader. LLMs were designed to imitate the very specific behavior of human writing; while they do this impressively, the underlying mechanisms of these models limit their capacities for meaning and naturalistic interaction, and their potential for dealing with the diversity in human language. We conclude by emphasising that LLMs are not theories of language, but tools that may be used to study language, and that can only be effectively applied with specific hypotheses to motivate research.

All talkers show some flexibility in their speech, and the ability to imitate an unfamiliar accent is a skill that shows vast individual differences. Yet the source of these individual differences, in particular whether they originate from perceptual, motor, or social/personality factors, is not yet clear. In the current study, we ask how individual differences in these factors predict individual differences in deliberate accent imitation. Participants imitated three accents, and attempts were rated for accuracy. A set of measures tracking individual differences in perceptual, motor, cognitive, personality, and demographic factors were also acquired. Imitation ability was related to differences in musical perception, vocal articulation, and the personality characteristic of "openness to experience," and was affected by attitudes towards the imitated talkers. Taken together, results suggest that deliberate accent imitation skill is modulated not only by core perceptual and motor skills, but also by personality and affinity to the talker, suggesting that some aspects of deliberate imitation are a function of domain-general constraints on perceptual-motor systems, while others may be modulated by social context.

Mate choice requires navigating an exploration-exploitation trade-off. Successful mate choice requires choosing partners who have preferred qualities; but time spent determining one partner's qualities could have been spent exploring for potentially superior alternatives. Here I argue that this dilemma can be modeled in a reinforcement learning framework as a multi-armed bandit problem. Moreover, using agent-based models and a sample of k = 522 real-world romantic dyads, I show that a reciprocity-weighted Thompson sampling algorithm performs well both in guiding mate search in noisy search environments and in reproducing the mate choices of real-world participants. These results provide a formal model of the understudied psychology of human mate search. They additionally offer implications for our understanding of person perception and mate choice.

Eye movements in the visual world paradigm are known to depend not only on linguistic input but on such factors as task, pragmatic context, affordances, etc. However, the degree to which eye movements may depend on task rather than on linguistic input is unclear. The present study for the first time tests how task constraints modulate eye movement behavior in the visual world paradigm by probing whether participants could refrain from looking at the referred image. Across two experiments with and without comprehension questions (total N = 159), we found that when participants were instructed to avoid looking at the referred images, the probability of fixating these reduced from 58% to 18% while comprehension scores remained high. Although language-mediated eye movements could not be suppressed fully, the degree of possible decoupling of eye movements from language processing suggests that participants can withdraw at least some looks from the referred images when needed. If they do so to different degrees in different experimental conditions, comparisons between conditions might be compromised. We discuss some cases where participants could adopt different viewing behaviors depending on the experimental condition, and provide some tentative ways to test for such differences.

What determines whether two people represent something in a similar way? We examined the role of verbal labels in promoting representational alignment. Across two experiments, three groups of participants sorted novel shapes from two visually dissimilar categories. Prior to sorting, participants in two of the groups were pre-exposed to the shapes using a simple visual matching task designed to reinforce the visual category structure. In one of these groups, participants additionally heard one of two nonsense category labels accompanying the shapes. Exposure to these redundant labels led people to represent the shapes in a more categorical way, which led to greater alignment between sorters. We found this effect of label-induced alignment despite the two categories being highly visually distinct and despite participants in both pre-exposure conditions receiving identical visual experience with the shapes. Experiment 2 replicated this basic result using more even more stringent testing conditions. The results hint at the possibly extensive role that labels may play in aligning people's mental representations.

Objects and places are foundational spatial domains represented in human symbolic expressions, like drawings, which show a prioritization of depicting small-scale object-shape information over the large-scale navigable place information in which objects are situated. Is there a similar object-over-place bias in language? Across six experiments, adults and 3- to 4-year-old children were asked either to extend a novel noun in a labeling phrase, to extend a novel noun in a prepositional phrase, or to simply match pictures. To dissociate specific object and place information from more general figure and ground information, participants either saw scenes with both place information (a room) and object information (a block in the room), or scenes with two kinds of object information that matched the figure-ground relations of the room and block by presenting an open container with a smaller block inside. While adults showed a specific object-over-place bias in both extending novel noun labels and matching, they did not show this bias in extending novel nouns following prepositions. Young children showed this bias in extending novel noun labels only. Spatial domains may thus confer specific and foundational biases for word learning that may change through development in a way that is similar to that of other word-learning biases about objects, like the shape bias. These results expand the symbolic scope of prior studies on object biases in drawing to object biases in language, and they expand the spatial domains of prior studies characterizing the language of objects and places.

Executing go/no-go or approach/avoidance responses toward a stimulus can change its evaluation. To explain these effects, some theoretical accounts propose that executing these responses inherently triggers affective reactions (i.e., action execution), while others posit that the evaluative influences originate from interpreting these responses as valenced actions (i.e., action interpretation). To test the role of action execution and action interpretation in these evaluative effects, we developed a novel training task that combined both go/no-go and approach/avoidance actions orthogonally. Participants either responded or did not respond (i.e., go/no-go) to control a shopping cart on screen, and as a result, either collected or did not collect (i.e., approach/avoidance) certain food items. When the task instructions referred to the go/no-go actions (Experiment 1, N = 148), we observed an effect of these actions. Participants evaluated no-go items less positively than both go and untrained items. No effect of approach/avoidance actions was observed. Contrarily, when the task instructions referred to the approach/avoidance actions (Experiment 2, N = 158), we observed an approach/avoidance effect. Participants evaluated approached items more positively and avoided items less positively than untrained items. No effect of go/no-go actions was observed. This suggests that action interpretation determined whether go/no-go or approach/avoidance actions influenced stimulus evaluation, when the same motor responses were made. Further examination of the role of action interpretation can inform theories of how actions influence stimulus evaluation, and facilitate the use of these interventions in applied settings.