Proceedings of the Annual Meeting of the Cognitive Science Society

The face inversion effect (FIE)—the disproportionate impairment in recognising inverted faces—has been central to debates between domain-specific and perceptual expertise accounts of visual processing. Although expertise-based explanations predict that inversion effects should emerge for non-face stimuli following learning, most supporting evidence derives from memory-based recognition tasks, leaving open the possibility that such effects reflect retrieval strategies rather than perceptual encoding. The present study tested this claim using a delayed matching-to-sample task with prototype-defined checkerboards, isolating perceptual encoding from long-term memory retrieval while manipulating stimulus complexity. Participants completed a categorisation training phase followed by a matching task involving familiar and novel stimuli presented upright or inverted. Results revealed a robust inversion effect for familiar stimuli only in the high-complexity condition, whereas no such effect emerged for simpler stimuli. These findings support the McLaren–Kaye–Mackintosh theory of perceptual learning by demonstrating that expertise modifies online perceptual encoding and that the engagement of holistic-like processing depends on stimulus complexity. We conclude that inversion effects reflect general expertise-based learning mechanisms operating under conditions of high informational load rather than face-specific perceptual architecture. Keywords: Perceptual expertise; Face inversion effect; Stimulus complexity; Perceptual learning; Visual categorisation; Holistic processing