We Trained an AI Paper Writer on Only Psychology Papers — Here's What It Produced

"Attachment theory, originally formulated by Bowlby (1969) and later expanded by Ainsworth's Strange Situation paradigm (Ainsworth et al., 1978), has evolved considerably in the past three decades. While early research focused primarily on infant-caregiver relationships, contemporary work has extended attachment principles to adult romantic relationships (Hazan & Shaver, 1987; Bartholomew & Horowitz, 1991) and psychotherapeutic contexts (Mallinckrodt et al., 2005). Longitudinal studies have demonstrated moderate stability in attachment patterns from infancy through adulthood (Waters et al., 2000), though significant life events can precipitate shifts between attachment categories (Fraley, 2002). Neurobiological investigations have identified distinct neural correlates for secure and insecure attachment, with secure attachment associated with enhanced prefrontal regulatory capabilities (Coan et al., 2006) and insecure patterns linked to hyperactivation of limbic structures during relationship stress (Gillath et al., 2005). Recent methodological advances have shifted from categorical to dimensional conceptualizations, with attachment now commonly measured along anxiety and avoidance dimensions, offering greater statistical power and theoretical precision (Fraley & Waller, 1998; Mikulincer & Shaver, 2007)..."

"Participants (N = 143; 76 female, 65 male, 2 non-binary; Mage = 19.7 years, SD = 1.2) were recruited from the university subject pool and received course credit for participation. Sample size was determined via a priori power analysis (G*Power; Faul et al., 2007) indicating 138 participants would provide 80% power to detect a medium effect (d = 0.5) with α = .05. Using a 2 (stereotype threat: present vs. absent) × 2 (mindset induction: fixed vs. growth) between-subjects design, participants were randomly assigned to conditions. In the stereotype threat condition, participants read that the upcoming test had previously shown gender differences in performance, while the no-threat condition described the test as gender-neutral. Mindset was manipulated using the procedure developed by Yeager et al. (2016), in which participants read articles about brain plasticity (growth condition) or brain specialization (fixed condition) and wrote a letter to a hypothetical struggling student. The dependent measure was performance on Raven's Advanced Progressive Matrices (RAPM; Raven et al., 1998), administered with a 20-minute time limit. After completing the RAPM, participants completed manipulation checks for stereotype threat awareness and mindset adoption, and were probed for suspicion using the funnel debriefing procedure (Bargh & Chartrand, 2000)..."

Prompt: Explain how source monitoring errors contribute to false memory formation.

"Source monitoring errors play a central role in false memory formation by compromising the cognitive processes that normally allow individuals to distinguish between different origins of mental experiences (Johnson et al., 1993). When encoding memories, we typically store both content information (what happened) and source information (how we acquired the knowledge). False memories often emerge when source monitoring—the set of processes for making attributions about memory origins—fails.

The Source Monitoring Framework (SMF) posits that source attributions rely on both rapid, heuristic judgments and more deliberate, systematic evaluations of memory characteristics (Johnson & Raye, 1981). False memories frequently occur when heuristic processing dominates and when memories from different sources share perceptual or semantic features. For instance, in the Deese-Roediger-McDermott (DRM) paradigm, presentation of semantically related words (bed, rest, awake) activates an associated but non-presented critical lure (sleep). During retrieval, activation of the critical lure can be misattributed to actual presentation rather than internal generation—a classic source monitoring error (Roediger & McDermott, 1995).

External suggestion can similarly induce source monitoring errors. In misinformation paradigms (Loftus, 2005), post-event information becomes incorporated into memory when individuals fail to correctly monitor the source of details as either from the original event or from subsequent suggestion. These errors are exacerbated when the suggested information is plausible and consistent with existing memory schemas (Brainerd & Reyna, 2002).

Neuroimaging evidence supports the SMF, showing that true memories typically elicit greater activation in sensory processing regions reflecting their perceptual encoding, while false memories show greater activation in conceptual processing regions, reflecting their generation through associative processes (Schacter et al., 1996). However, when false memories are accompanied by vivid perceptual details—a phenomenon termed 'phantom recollection' by Brainerd et al. (2001)—the neural activation patterns can closely resemble those of true memories, making them particularly compelling and resistant to correction..."