Endogenous attention biases transformational apparent motion based on high-level shape representations

Abstract

When two pre-existing, separated squares are connected by the sudden onset of a bar between them, viewers do not perceive the bar to appear all at once. Instead, they see an illusory morphing of the original squares over time. The direction of this transformational apparent motion (TAM) can be influenced by endogenous attention deployed before the appearance of the connecting bar. Here, we investigated whether the influence of endogenous attention on TAM results from operations over high-level feature-independent shape representations, or instead over lower level shape representations defined by specific visual features. To do so, we tested the influence of endogenous attention on TAM in first- and second-order displays, which shared common shapes but had different shape-defining attributes (luminance and texture contrast, respectively). In terms of both the magnitude of directional bias and timing, we found that endogenous attention exerted a similar influence on both first- and second-order objects. These results imply that endogenous attention biases the perceived direction of TAM by operating on high-level shape representations that are invariant to the low-level visual features that define them. Our results support a four-stage model of TAM, where a feature encoding stage passes a features-specific layout to a parsing stage that forms discrete, high-level meta-featural shapes, which are then matched and visually interpolated over time.

Figure 1.

Transformational apparent motion. When a bar appears between adjacent, identical squares (a), participants perceive both squares as continuously changing shape—appearing to extend toward each other and collide in the middle (b). If attention is directed to one of the squares, motion is perceived away from the attended square (c). Yellow arrows indicate the direction the perceived motion.

Figure 2.

Four-stage model of TAM processing. (I) Encoding: low-level features and feature-specific shapes are detected at an early stage. Attention (red outlines) modulates first-order and second-order features differently (e.g., Allen & Ledgeway, 2003). The resulting activity is fed forward to the sequence of parsing, matching, and motion interpolation operations that result in TAM. (II) Parsing: meta-featural shapes are extracted from the input. Low-level feature information is discarded from the stimuli and feature specific attentional effects from the encoding stage (if any) may or may not carry over. The output of the parser is an abstract feature-independent shape representation. (III) Matching: shape contours that correspond with each other are matched across the two time intervals, t₁ and t₂. At this stage, the direction of motion could be biased by attentional differences (top and middle rows), or it could remain ambiguous in the absence of any attentional cues (bottom row). (IV) Motion interpolation: the motion that must have occurred in the world to give rise to the image sequence is reconstructed based on the correspondence between shapes.

Figure 3.

Square, arrow, and bar stimuli. At the start of each trial, participants saw four squares with a side length of 3.92°, positioned 11.40° away from a central arrow cue and 16.07° from each other, adjacently (a, first order; c, second order). Then, they saw an additional stimulus appear: a long bar bridging adjacent squares, 16.07° length (b, first order; d, second order). Note that the drawings in yellow do not reflect what participants saw, but have been added for illustration purposes.

Figure 4.

Experimental paradigm. Both first-order (left, a–c) and second-order (right, d–f) trials followed the same procedure. In experimental trials, we presented a central fixation dot (red), then a display of four squares with a central cue arrow, then a connecting bar between two adjacent squares (a, d). In incremental-motion catch trials, we presented a connecting bar incrementally, extending from an adjacent square toward the cued square (b, e). In invalid catch trials, we presented a connecting bar between two uncued squares (c, f). At the end of each trial, participants pressed one of four arrow keys indicating the direction of perceived motion. Note that yellow lines in (d–f) do not reflect what participants saw but have been added to illustrate the edges of second-order, texture-defined objects. SOA, stimulus onset asynchrony.

Figure 5.

Proportion of experimental trials with congruent motion percept, by stimulus onset asynchrony (SOA). At longer SOAs (≥250 ms), participants perceived motion away from the cued square (congruent) in a significantly higher proportion of trials than would be expected by chance (50%). This was true for both first-order stimuli (blue) and second-order (red) stimuli. Each SOA at which participants perceived motion in a particular cue-relative direction in more than half of trials (significant at p < 0.05) is denoted with an asterisk. Standard error of the mean for each stimulus type, at each SOA, is shown in black.

Figure 6.

Proportion of congruent percepts in experimental versus real-motion trials, at longest stimulus onset asynchrony (SOA). At the longest SOAs (≥250 ms), where participants perceived congruent motion in the majority of trials, they perceived congruent motion comparably for first- and second-order displays. In both cases, congruent motion was perceived in a significantly higher proportion of real-motion catch trials versus experimental transformational apparent motion (TAM) trials.