Shared variance across FrACT visual tasks: Robust positive associations but inconclusive latent structure

Visual performance was positively associated across eight low-level acuity and hyperacuity tasks in an adult study, with the strongest relationships emerging between conceptually similar measures. The study found that 27 of 28 pairwise correlations survived false-discovery-rate correction, and Bayesian analyses supplied strong evidence for nonzero associations in most task pairs. However, exploratory factor analyses did not produce a stable model of the shared variation, leaving the existence of a single general visual ability unresolved.

Researchers tested 158 adults using eight paradigms implemented in FrACT10: single letter, Landolt C, tumbling E, row of optotypes, flanking rings, frame ring, Vernier two-line and Vernier three-line tasks. After outlier exclusion, the main analyses included 132 participants. Within-session test–retest reliability was moderate for most of the tasks. The investigators asked whether individual differences in visual performance become more evident when the measures are similar in both concept and procedure, addressing a long-standing question about whether performance reflects a general visual ability or a collection of largely task-specific processes.

The pairwise results formed what the authors described as a broad positive manifold. Every correlation was positive, and 27 of 28 remained significant after correction for false discovery rate. Bayesian analyses also provided strong evidence for nonzero associations in most task pairs, although association strength varied markedly with task similarity. Relationships were robust among optotype-based tasks, while links between optotype and Vernier paradigms were weaker. The pattern shows that the performance measures were not wholly independent, but it also indicates that the overlap between them depended on which tasks were included in the battery.

Exploratory factor analyses indicated substantial shared covariance among the tasks but failed to identify a dependable latent structure. A one-factor model fit poorly, while two-factor and three-factor solutions produced improper estimates that made them difficult to interpret. The researchers therefore supported an intermediate account: visual individual differences appeared neither fully independent across tasks nor reducible to one strong general factor. Instead, shared variance was partial and task-dependent, meaning that conclusions about general visual ability could vary with the composition of the test battery. The study was not preregistered, and the inconclusive factor models limit what can be concluded about the number or structure of the underlying factors. Taken together, the positive correlations and their dependence on task similarity suggest that closely matched visual tasks can capture overlapping aspects of performance, while the data do not establish a single, stable general visual trait.

Subject of Research: Individual differences and shared variance across eight FrACT10 visual acuity and hyperacuity tasks

Article Title: Shared variance across FrACT visual tasks: Robust positive associations but inconclusive latent structure

Article References: Carabba, F., Altieri, E., Ghiani, A., Musa, M., Rosengart, E., Maniglia, M., & Battaglini, L. (2026). Shared variance across FrACT visual tasks: Robust positive associations but inconclusive latent structure. Attention, Perception, & Psychophysics, 88(8), Article 208. https://doi.org/10.3758/s13414-026-03353-1

Image Credits: AI Generated

DOI: 10.3758/s13414-026-03353-1

Keywords: visual perception, individual differences, task similarity, acuity, hyperacuity, latent structure, factor analysis, test–retest reliability, Shared, variance, across, FrACT