Source
Ivy & Mroczko-Wąsowicz (2025), Framing Effects in Object Perception, LIT-099, read in full in NOTE-150 (Review of Philosophy and Psychology 16:969–996).
What was actually shown
The definition (p. 975). “A perceptual frame is an adaptation of the perceptual system caused by perceptual learning and realized through bottom-up functional processes such that sensory information is organized in a subject-dependent way leading to idiosyncratic perceptual object representations.” Frames are defined against attention, which is “universally accessible to any perceiver with normally functioning capacities” (p. 974); a frame needs training. Simple objects, such as a single dot in a lattice, are individuated by any normal perceiver, so the thesis covers a class of objects only.
The separation (§4, pp. 989–991). Cognition has “a direct and unmediated influence on perception insofar as it is through top-down influences that perceptual frames are encoded” (p. 990). Once installed, a frame needs no occurrent cognitive input: “the cognitive origin that explains why a perceptual frame was implemented in the first place need not be the same reason why that frame operates in early sensory processing”. Whether individual episodes are penetrable “remains an open question” (p. 990), and frames are compatible with penetration (fn. 2).
The evidence offered. Reliability-weighted cue integration with learned priors (Ernst & Banks 2002; Helbig & Ernst 2007), prism-adaptation after-effects, early amodal completion (initial responses about 140 ms, differential about 240 ms), Teufel et al.’s (2018) two-tone images that cannot be unseen, expert anticipation, and radiologists at about 200 ms per image. The only first-party data is the authors’ own eye-tracking study (Ivy et al. 2023), cited for experts applying domain search patterns outside their domain (p. 976).
What could have come out otherwise. Nothing in the paper. It runs and
proposes no test, and NOTE-150 grades its claims weak to moderate. What
would test it is in promote_when.
What this does not say
- That frames have been shown to differ from attention or imagery. No cited study separates the accounts, and the argument against imagery rests on one study (NOTE-150, Limitations).
- That frames operate bottom-up in any clear sense. §3.iii sums up its evidence as the effect of “having applied a perceptual frame (a specific top-down prediction or prior knowledge)” (p. 988). On a predictive-processing reading, priors are applied top-down on every episode, and the paper does not say how that squares with its definition.
- That different frames yield different objects. The multisensory evidence shows different weightings and estimates, for example of bar height. “Different objects” is inferred.
- That perception is not penetrable. The paper leaves the episode-level question open. The seeded skim’s “without penetrating any single percept” was corrected by NOTE-150 to “whether or not”.
- That modularists are refuted. The verdict that Firestone & Scholl’s formulation is “inaccurate” reads “direct, unmediated” diachronically. Perceptual learning is an effect they set aside as not penetration.
- That every completion needs a frame. The three-step argument of p. 983 states the converse of its premise; its conclusion that completion needs a reason survives, but that the reason is always a frame is stipulated (“where there is ‘prior knowledge,’ there is a perceptual frame at work”, p. 980).
Connections
- The paper presupposes a theory of representation and states none. NOTE-150 names Martínez (THEORY-529) as one it could use.
- Cross-modal integration and unity (LIT-190) is the record’s nearest neighbour on multisensory binding, about consciousness rather than object individuation.
Note of 2026-10-29: what holds it back
Kept Proposed by the sixteenth status review. NOTE-150 read the paper in full, and the paper runs no test. No experiment in the record separates a frame from attention or imagery with the individuated object as the measure, and the criterion promote_when asks for first, saying which learned tunings are frames, is not stated. The evidence it rereads shows changed weightings and estimates, not different objects.