Where is mirror invariance? Masked priming effects by mirrored and rotated transformations of reversible and nonreversible letters

• Reversible letters include mirror images (d - b) and plane rotations (d - p). • We investigated the influence of these orientation transformations of letters with masked priming. • Lexical decision (Exp 1 & 2) and same-different task on words and letters (Exp 3 & 4). • Mirrored and rotated letters led to slower performance than identical letters. • Larger orientation cost for reversible than nonreversible letters. • Across experiments mirror invariance not found during orthographic processing. The cognitive mechanism underpinning mirror priming effects for letters (e.g., IDEA primed by ibea) is still unclear: it could be about mirror invariance (perceptual equivalence of mirror images - d and b - which does not extend to other orientation contrasts like plane rotations - d and p) or about visual similarity (b, d, and p share the same visual features and shape). To enlighten it and investigate the modulatory role of task and orthographic context (words vs single letters), we manipulated letter (nonreversible; reversible) and prime condition (identity, mirrored, rotated, control) in four masked priming experiments: lexical decision with conventional and sandwich paradigms (Experiments 1 and 2), same-different task on words and single letters (Experiments 3 and 4). All experiments replicated the differential mirror priming effect: inhibitory for reversible letters, facilitatory for nonreversible letters. Notably, similar priming effects were found for plane rotations. Regardless of task or orthographic context, a reliably larger orientation cost was found for reversible than nonreversible letters for both mirror images and plane rotations. These results suggest that the underlying mechanism is not mirror invariance, but rather visual similarity allied with prelexical dynamics.