First- and second-order transformational apparent motion rely on common shape representations

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

“…Next, we explored motion percepts in invalidly cued catch trials ( Figures 4 c, 4 f) at the SOAs of interest. This measure was included to ensure that participants were not favorably perceiving or reporting motion in a particular direction in the absence of attentional influences (i.e., when they were not volitionally attending either of the squares connected by the bar stimulus, as in Hartstein et al 2021 ). We found that participants did not significantly favor a particular direction in invalidly cued catch trials where the direction of motion was vertical, for both first- and second-order displays.…”

“…Second-order stimuli were defined by dynamically updating black and white textures on a statically presented black and white background ( Figures 4 d–f). In second-order trials, black and white background and stimulus textures were created using the same procedure as described in Hartstein et al. (2021) .…”

“…Several studies have provided compelling evidence that TAM and its variants are high-level phenomena that involve an inference of a change in shape, regardless of how that shape was defined in the stimulus. For example, Hartstein et al. (2021) showed that the direction of motion was determined by shape correspondence between stimuli in a similar manner for both first- and second-order displays, indicating that TAM is invariant to low-level visual cues.…”

“…These instances of illusory motion driven by shape cannot be explained by the prior entry account or attentional gradients. We have, therefore, suggested that TAM is instead driven by the parsing and matching of the successive shapes presented in the image sequence ( Hartstein, Saleki, Ziman, Cavanagh, & Tse, 2021 ; Tse et al., 1998 ). These operations decide which of the figures had to change shape to account for their new arrangement.…”

“…Here, we compared first- and second-order TAM in a case where the direction of motion was influenced solely by endogenous attention. We predicted that the effects of endogenous attention on initial shape extraction (featural level) would be lost once the shapes were encoded into a meta-featural representation, in the sense that it defines a figure via multiple possible shape-defining features, known to be used for the parsing and matching stages of TAM ( Hartstein et al., 2021 ). Therefore, the high-level representations of the second-order stimuli would show the same perceived direction as that seen for first-order stimuli across all stimulus onset asynchronies (SOAs).…”