Abstract

Studies of the mediating factors of imaginary companions (ICs) in children and adults are well-represented in the literature. However, the nature and structure of behavioral expressions in IC characters have been less formally scrutinized. We examined these issues in a convenience sample of 389 adults. Of these, 155 self-reported childhood ICs and retrospectively characterized their IC phenomenology via a set of 14 items modified from previous research. Rasch analyses showed that IC experiences form a true probabilistic hierarchy whose structure varied little across respondents’ age, gender, having siblings, as well as the respondents’ number of past ICs, or their decision to inform others about their IC. This hierarchy starts with shallow (i.e., ostensibly ‘adaptive or positive’) experiences and transition to deep (i.e., relatively ‘maladaptive or negative’) contents at higher levels. Network analysis suggested that respondents invented ICs primarily to combat feelings of loneliness. However, contrary to the Rasch model, when comparing shallow vs deep IC experients’ answers, positive and negative perceptual contents lost their distinction, thereby severely distorting measurement. These distortions were sufficiently powerful to reliably predict respondents’ group membership. Results derived from retrospective accounts of childhood experiences, which might differ from IC contents and dynamics measured either in real-time or within adult populations. However, these findings suggest that ICs comprise a latent dimension of shallow-to-deep perceptions that might relate to schizotypal or dissociative phenomena manifesting in everyday contexts.