Can infants adopt underspecified contents into attributed beliefs? Representational prerequisites of theory of mind

Do early meanings of negation map onto a fully-fledged negation concept in infancy?

Young children acquire an amazing knowledge base, rapidly learning from, and even going beyond the observable evidence. They arrive at forming abstract concepts and generalizations and recruit logical operations. The question whether young infants can already rely on abstract logical operations, such as disjunction or negation, or whether these operations emerge gradually over development has recently become a central topic of interest. Here we target this question by focusing on infants' early understanding of negation. According to one view, negation comprehension is initially restricted to a narrow range of meanings (such as rejection or non-existence) and only much later infants develop a broader understanding that maps onto a fully-fledged negation concept. Alternatively, however, infants may rely on a fully-fledged negation concept from early on, but some forms of negation may pose more mapping and processing difficulties than others. Here we tested infants' understanding of two syntactically and semantically different forms of negation, existential negation and propositional denial in a language (Hungarian) that has a separate negative particle for each, and thus the two negation forms can be directly compared. We engaged 15- and 18-month-old infants in a search task where they had to find a toy in one out of two locations based on verbal utterances referring to the object at one of the locations involving existential negation (Nincsen - not.be.3SG) or propositional denial (Nem itt van - not here be.3SG). In Experiments 1-3 we found a parallel development for these two kinds of negation. 18-month-olds successfully comprehended both, while 15-month-olds were at chance for both. In Experiment 4 we excluded the possibility that 15-month-olds' chance performance is explained by task-related difficulties, as they succeeded in a similar, but nonverbal task. Thus, 15-month-olds likely still have not solved the mapping for the two negation forms. The parallel performance of the two age groups with the two negation types (failing or succeeding on both) is consistent with the hypothesis that different forms of negation rely on similar conceptual underpinnings already in early development.

Testing theory of mind in large language models and humans

At the core of what defines us as humans is the concept of theory of mind: the ability to track other people's mental states. The recent development of large language models (LLMs) such as ChatGPT has led to intense debate about the possibility that these models exhibit behaviour that is indistinguishable from human behaviour in theory of mind tasks. Here we compare human and LLM performance on a comprehensive battery of measurements that aim to measure different theory of mind abilities, from understanding false beliefs to interpreting indirect requests and recognizing irony and faux pas. We tested two families of LLMs (GPT and LLaMA2) repeatedly against these measures and compared their performance with those from a sample of 1,907 human participants. Across the battery of theory of mind tests, we found that GPT-4 models performed at, or even sometimes above, human levels at identifying indirect requests, false beliefs and misdirection, but struggled with detecting faux pas. Faux pas, however, was the only test where LLaMA2 outperformed humans. Follow-up manipulations of the belief likelihood revealed that the superiority of LLaMA2 was illusory, possibly reflecting a bias towards attributing ignorance. By contrast, the poor performance of GPT originated from a hyperconservative approach towards committing to conclusions rather than from a genuine failure of inference. These findings not only demonstrate that LLMs exhibit behaviour that is consistent with the outputs of mentalistic inference in humans but also highlight the importance of systematic testing to ensure a non-superficial comparison between human and artificial intelligences.

Meaning as mentalization

The way we establish meaning has been a profound question not only in language research but in developmental science as well. The relation between linguistic form and content has been loosened up in recent pragmatic approaches to communication, showing that code-based models of language comprehension must be augmented by context-sensitive, pragmatic-inferential mechanisms to recover the speaker's intended meaning. Language acquisition has traditionally been thought to involve building a mental lexicon and extracting syntactic rules from noisy linguistic input, while communicative-pragmatic inferences have also been argued to be indispensable. Recent research findings exploring the electrophysiological indicator of semantic processing, the N400, have raised serious questions about the traditional separation between semantic decoding and pragmatic inferential processes. The N400 appears to be sensitive to mentalization-the ability to attribute beliefs to social partners-already from its developmental onset. This finding raises the possibility that mentalization may not simply contribute to pragmatic inferences that enrich linguistic decoding processes but that the semantic system may be functioning in a fundamentally mentalistic manner. The present review first summarizes the key contributions of pragmatic models of communication to language comprehension. Then, it provides an overview of how communicative intentions are interpreted in developmental theories of communication, with a special emphasis on mentalization. Next, it discusses the sensitivity of infants to the information-transmitting potential of language, their ability to pick up its code-like features, and their capacity to track language comprehension of social partners using mentalization. In conclusion, I argue that the recovery of meaning during linguistic communication is not adequately modeled as a process of code-based semantic retrieval complemented by pragmatic inferences. Instead, the semantic system may establish meaning, as intended, during language comprehension and acquisition through mentalistic attribution of content to communicative partners.

Children's use of reasoning by exclusion to infer objects' identities in working memory

Reasoning by exclusion allows us to form more complete representations of our environments, "filling in" inaccessible information by ruling out known alternatives. In two experiments (Experiment 1: N = 34 4- to 6-year-olds; Experiment 2: N = 85 4- to 8-year-olds), we examined children's ability to use reasoning by exclusion to infer the identity of an unknown object and investigated the role of working memory in this ability. Children were asked to encode a set of objects that were then hidden, and after a brief retention interval children were asked to select the identity of the object hidden in one of the locations from two alternatives. On some trials, all the images were visible during encoding, so selecting the correct identity when probed required successful working memory storage and retrieval. On other trials, all but one of the images was visible during encoding, so selecting the correct identity when probed also required maintaining a representation of an unknown object in working memory and then using reasoning by exclusion to fill in the missing information retroactively to complete that representation by ruling out known alternatives. To investigate the working memory cost of exclusive reasoning, we manipulated the working memory demands of the task. Our results suggest that children can use reasoning by exclusion to retroactively assign an identity to an incomplete object representation at least by 4 years of age but that this ability incurs some cognitive cost, which eases with development. These results provide new insights into children's representational capacities and on the foundational building blocks of fully developed exclusive reasoning.

Task-independent neural bases of peer presence effect on cognition in children and adults

There is ample behavioral evidence that others' mere presence can affect any behavior in human and non-human animals, generally facilitating the expression of mastered responses while impairing the acquisition of novel ones. Much less is known about i) how the brain orchestrates the modulation of such a wide array of behaviors by others' presence and ii) when these neural underpinnings mature during development. To address these issues, fMRI data were collected in children and adults alternately observed and unobserved by a familiar peer. Subjects performed a numerosity comparison task and a phonological comparison task. While the former involves number-processing brain areas, the latter involves language-processing areas. Consistent with previous behavioral findings, adults' and children's performance improved in both tasks when observed by a peer. Across all participants, task-specific brain regions showed no reliable change in activity under peer observation. Rather, we found task-independent changes in domain-general brain regions typically involved in mentalizing, reward, and attention. Bayesian analyses singled out the attention network as the exception to the close child-adult resemblance of peer observation neural substrates. These findings suggest that i) social facilitation of some human education-related skills is primarily orchestrated by domain-general brain networks, rather than by task-selective substrates, and ii) apart from attention, peer presence neural processing is largely mature in children.

Spontaneous attribution of underspecified belief of social partners facilitates processing shared information

The main question of Theory of Mind research is not only how we represent others' mental states, but also how these representations influence our first-person interaction with our surrounding environment. A novel theory of belief files proposes that we should think about belief tracking as an online, spontaneous, and effortless mechanism giving rise to structured representations, thus easing the use of beliefs in behavior selection. Beliefs are formed by two different sub mechanisms: (1) opening an empty placeholder belief file, for a particular intentional agent, and (2) filling it up with mental content attributed to the agent. This theory opens the possibility of exploiting theory of mind abilities even in situations when we can attribute only underspecified mental contents to others. The goal of the present study was to provide a proof of concept test: whether spontaneous belief tracking starts effortlessly even when we do not know a partner's actual belief content. We created an object detection paradigm, where the visual access of a virtual agent to the object to be detected by the participant was manipulated. The agent getting access to the information for processing always preceded the participant getting access to it, resulting in the need of attributing belief without specified content in it. Our results have shown that participants detected the object with a reduced reaction time when the observed agent had visual access to the object's expected place compared to when the participant watched the same scenario, but the object's location remained occluded for the observed agent and thus was revealed only for the participant. This suggests that the information processing of humans speeds up when another agent has access to a piece of information as well. Thus, we do track agents' potential beliefs without knowing its actual content. This study contributes to our understanding of the effect of spontaneous computation of others' mental states on first-person information processing.

Seeing the World From Others' Perspective: 14-Month-Olds Show Altercentric Modulation Effects by Others' Beliefs

Humans have a propensity to readily adopt others' perspective, which often influences their behavior even when it seemingly should not. This altercentric influence has been widely studied in adults, yet we lack an understanding of its ontogenetic origins. The current studies investigated whether 14-month-olds' search in a box for potential objects is modulated by another person's belief about the box's content. We varied the person's potential belief such that in her presence/absence an object was removed, added, or exchanged for another, leading to her true/false belief about the object's presence (Experiment 1, n = 96); or transformed into another object, leading to her true/false belief about the object's identity (i.e., the objects represented under a specific aspect, Experiment 2, n = 32). Infants searched longer if the other person believed that an object remained in the box, showing an altercentric influence early in development. These results suggest that infants spontaneously represent others' beliefs involving multiple objects and raise the possibility that infants can appreciate that others encode the world under a unique aspect.

Beliefs and self-reported practice of footcare among persons with type II diabetes mellitus attending selected health centres in east Trinidad

Background

It has been documented that nearly 600 million people worldwide are expected to have diabetes mellitus in 2035 and that approximately 140,000 persons aged 20-69 years living with diabetes mellitus in Trinidad and Tobago. It is also indicated that patients with type II diabetes mellitus face a higher risk of diabetic foot neuropathy and foot ulcers which increase the risk of below the knee amputation in persons living with diabetes.

Purpose

The aim of this research project was to explore the beliefs related to footcare and the self-reported footcare practice of persons with type II diabetes mellitus attending selected health centres in East Trinidad.

Method

A survey was used to explore the footcare beliefs and the self-reported footcare practice of persons with type II diabetes mellitus attending selected health centres in East Trinidad (n = 87).

Results

Participants had strong belief regarding susceptibility to foot injury, strong belief regarding the seriousness of complications of foot injury, reported good footcare practice and excellent practice of overcoming barriers to performing footcare. There is a correlation between the belief regarding susceptibility to a foot injury and age (p ≤ 0.05). Also, there is a significant correlation between belief regarding susceptibility to a foot injury, seriousness of complications and self-reported footcare practices (p ≤ 0.05).

Conclusion

This study explores and describes the beliefs and self-reported practices of footcare among individuals with type II diabetes mellitus at selected health centres in East Trinidad. It supports the Health Belief Model as an effective framework for the promotion of appropriate footcare among persons with type II diabetes mellitus.