Patterns of long-term change in implicit social cognition

Abstract

To navigate the social world, humans have developed the adaptive capacity to rapidly form attitudes and beliefs about the many people and social groups that make up their environments. Many of these attitudes and beliefs are implicit, arising automatically in mind upon encountering a member of a social group, and subsequently guiding consequential social behaviors such as whom to approach or avoid. However, to be fully adaptive, implicit attitudes and beliefs must also be capable of change. As history unfolds to reveal new distributions of social roles, or new representations of social groups in the media, so too should the implicit attitudes and beliefs about those groups update in response. Although it has been shown that implicit attitudes and beliefs are temporarily malleable (on the short-term, within a single experimental session), evidence of durable, long-term changes in implicit attitudes and beliefs has remained elusive. To that end, the current dissertation presents the first comprehensive evidence of long-term change in implicit social cognition, providing a record from which to generate new predictions on why, how, and for whom implicit attitudes and beliefs reveal durable change. Part I (Charlesworth & Banaji, 2019, Psychological Science) uncovers patterns of long-term change and stability in implicit attitudes towards six social groups (age, disability, body weight, race, skin-tone, and sexual orientation), using time series analyses applied to data from over 4.4 million tests of implicit attitudes collected continuously for an entire decade (2007-2016). Additional results from a further 1.5 million tests of two implicit gender stereotypes (male-career/female-family and male-science/female-arts; Charlesworth & Banaji, in press, Social Psychological and Personality Science) are reported to expand the scope of understanding in long-term implicit social cognition change. Part II (Charlesworth & Banaji, in press, American Psychologist) probes deeper into these patterns to examine whether the observed long-term trends in implicit attitudes are (a) widespread and unfolding in parallel across multiple demographic groups in society or (b) idiosyncratic and non-parallel across specific demographic groups. Finally, Part III (Charlesworth et al, 2021, Psychological Science) demonstrates a new method for measuring social cognition at unprecedented scales, using natural language processing techniques (word embeddings) to identify gender stereotypes in over 65 million words of natural child and adult language. Such methods have the potential to examine implicit attitude and belief change at timescales of two hundred years or more. Ultimately, the research presented in this dissertation illustrates the methods, empirical records and initial hypotheses that can be used to guide a new understanding of the nature of implicit social cognition change and the dynamic interweaving between mind and society.