Robert Lee, Ian Apperly, Rory Devine and Sanne Van Der Kleij
Affiliation: University of Birmingham
Category: Psychology
Keywords: Psychometrics
Date: Wednesday 3rd of September
Time: 15:00
Location: Room 154 (154)
View the full session: Mind Reading
Introduction
One of the most important foundations of the ability to build and maintain social relationships among children and younger adolescents is the ability to understand the perspectives, thoughts, and feelings of other people, which we refer to as âmindreadingâ. We predict that mindreading is a key part of âsocial individual differencesâ that support social competence amongst young people (Apperly, 2012), with the consequence that if school and workplace environments support young people to develop their mindreading skills, this would have the potential to support their overall social competence.
Limitations of existing research
Mindreading abilities continue to develop across middle childhood and adolescence (e.g. Devine et al., 2024) and individual differences in effective mindreading remain relatively stable across this developmental period (Devine, 2021). However, most existing measures of mindreading are incapable of measuring mindreading reliably in older adolescents and adults (primarily due to ceiling effects) and no existing measure provides consistent psychometric evidence of reliability and validity (Yeung et al., 2024). Consequently, whilst it is likely that there is continuing variation in advanced mindreading abilities amongst the adult population, research into such variation and the cognitive abilities that underpin it is hampered by a lack of robust measures that are sensitive to individual differences in advanced mindreading ability in adolescence and adulthood.
It is also the case that current theories of mindreading struggle to explain how such abilities continue to vary beyond the age when children appear to have all of the concepts considered necessary for basic mindreading. Developing a robust measure of advanced mindreading will support us to further our understanding of how and why mindreading abilities continue to vary amongst adults. Our working hypothesis is that mindreading continues to vary because some people are better at adapting their mindreading with the flexibility necessary to understand a wide range of different people and situations.
Development of a new advanced mindreading task
To address this challenge, we have developed a task designed to measure advanced mindreading in young people aged 15-30. The task was crowd-sourced through participatory research in which real social narratives written by a diverse range of young people in schools and colleges across Birmingham, UK. These story authors set and answered questions that explored their motivations within their story, which formed the âground truthâ in our coding manual for what that person was thinking or feeling.
We selected 25 stories from our pool of 75, then ran a pilot study (N=300). Pre-registered analyses indicated 10 viable story-question items, each with 1 simple comprehension question, and 1 associated mindreading question, with open-text responses. To test the validity and reliability of our new measure, we then recruited a large sample of school and college students (N=660 ages 13-18), plus online participants (N=1952, ages 18-30).
Findings
Latent variable modelling was used to examine the latent factor structure of the mindreading task, using Confirmatory Factor Analysis (CFA) in our best-fitting model from our pilot study, across a discovery sample (N=1306) and validation sample (N=1306). This indicated that participant responses to 9 stories loaded onto a single latent factor. Using the whole dataset (N=2612), further CFA examined the extent to which the latent factor explained performance on each item, by regressing each item onto each participantâs score for average response length, and the average percentage of mental state terms they used. Results indicate that the measure remained psychometrically robust to the potentially confounding effects of response length, and also to the proportion of mental state terms used by participants in their responses. This suggests that performance on the mindreading task was primarily determined by accurate mindreading inferences.
Further CFAs with covariates were used to examine the convergent and discriminant validity of the new mindreading measure. Supporting the validity of the novel task, these tests indicated that performance on the mindreading task correlates moderately and positively with performance on an existing measure of advanced mindreading (the Movie Assessment of Social Cognition, MASC, Dziobek et al., 2006). The correlation between these tasks was stronger than that between the mindreading task and a measure of general cognitive ability. Additionally, performance on the mindreading task showed a unique positive association with the MASC when controlling for the combined impact of individual differences in age, gender, socio-economic background, and general cognitive ability.
Implications
These tests indicate that the novel mindreading task measures the same construct as one of the few other measures of advanced mindreading that currently exists and that performance cannot be reduced to general cognitive ability. Our measure shows convergent and discriminant validity. Since it is based on the social understanding of a diverse range of story authors, our task is more likely than existing measures to capture the everyday variation in mindreading across different people and situations. We will discuss how this work can inform new theory capable of explaining the existence of variation in mindreading in adults.