Line orientation adaptation: local or global?

Abstract

Prolonged exposure to an oriented line shifts the perceived orientation of a subsequently observed line in the opposite direction, a phenomenon known as the tilt aftereffect (TAE). Here we consider whether the TAE for line stimuli is mediated by a mechanism that integrates the local parts of the line into a single global entity prior to the site of adaptation, or the result of the sum of local TAEs acting separately on the parts of the line. To test between these two alternatives we used the fact the TAE transfers almost completely across luminance contrast polarity [1]. We measured the TAE using adaptor and test lines that (1) either alternated in luminance polarity or were of a single polarity, and (2) either alternated in local orientation or were of a single orientation. We reasoned that if the TAE was agnostic to luminance polarity and was parts-based, we should obtain large TAEs using alternating-polarity adaptors with single-polarity tests. However we found that (i) TAEs using one-alternating-polarity adaptors with all-white tests were relatively small, increased slightly for two-alternating-polarity adaptors, and were largest with all-white or all-black adaptors. (ii) however TAEs were relatively large when the test was one-alternating polarity, irrespective of the adaptor type. (iii) The results with orientation closely mirrored those obtained with polarity with the difference that the TAE transfer across orthogonal orientations was weak. Taken together, our results demonstrate that the TAE for lines is mediated by a global shape mechanism that integrates the parts of lines into whole prior to the site of orientation adaptation. The asymmetry in the magnitude of TAE depending on whether the alternating-polarity lines was the adaptor or test can be explained by an imbalance in the population of neurons sensitive to 1(st)-and 2(nd)-order lines, with the 2(nd)-order lines being encoded by a subset of the mechanisms sensitive to 1(st)-order lines.

Figure 1. Stimuli used in the experiments.

One can experience the tilt after-effect (TAE) obtained with single polarity line adaptors (A) and one-alternating polarity line (B) by moving one’s eyes back and forth along the markers located midway between the pair of adapting lines (left) for about 60s, and then shifting one’s gaze to the middle of the single single polarity line test (right). (C) Example lines made of either all white (W), all dark (D), one-alternating polarity (1A) and two-alternating polarity (2A) micropatterns. (D) Example lines in which the Gabor patches were either all tangentially oriented to the line’s path, termed ‘snake’ (S), all orthogonally oriented to the path, termed ‘ladder’ (L), alternating in orientation every other element (1AO) or every two consecutive elements (2AO). (e) Schematic representation of the adapting and test procedure - see text for details.

Figure 2. Results for Experiment 1:

TAEs induced in all-white test lines (A) and one-alternating polarity test lines (B), by all-white (white bars), all-dark (black bars), one-alternating (light gray bars) and two-alternating polarity (dark gray bars) line adaptors, for each observer and the average across observers. (C–D) Normalized TAE obtained for each ‘different’ adaptor-and-test condition to the after-effect obtained using the ‘same’ adaptor-and-test condition, for each observer and the average across observers. One can think of this measure as the amount of transfer of the after-effect in the ‘different’ condition. Transfer of TAE obtained in the different adaptor-and-test condition normalized to the after-effect obtained with (C) all-white adaptor-and-test condition and (D) one-alternating polarity adaptor-and test condition. A transfer value of 1 (dashed lines) indicates that the after-effects obtained in the ‘different’ and ‘same’ adaptor/test conditions are similar in magnitude.

Figure 3. Results for Experiment 2:

TAEs induced in snake (A) and one-alternating-orientation test lines (B) by snake (white), ladder (black), one-alternating (dense hatched) and two-alternating-orientation (sparse hatched) adaptors, for each observer and the average across observers. (C–D) Transfer of TAE obtained in the different adaptor-and-test condition normalized to the after-effect obtained with (C) snake adaptor-and-test condition and (D) one-alternating polarity adaptor-and test condition, for each observer and the average across observers. A transfer value of 1 (dashed lines) indicates that the after-effects obtained in the ‘different’ and ‘same’ adaptor/test conditions are similar in magnitude.

Figure 4. Example parts-jittered lines used in Experiment 3.

One can experience the TAE obtained with (A) jittered one-alternating polarity line adaptor for large (right), small (center) and no (left) jitter condition and no jittered one-alternating polarity line test. (B) Example parts-jittered white lines for large (right), small (center) and no (left) jitter condition.

Figure 5. Results for Experiment 3– polarity condition.

TAEs obtained with (A) all white lines adaptor-and-test for small (light gray bars), large (dark gray bars) and no jitter (white bars) adaptor conditions, for each observer and the average across observers. (B) one-alternating polarity adaptor-and-test lines for small (hatched light gray bars), large (hatched dark gray bars) and no jitter (hatched white bars) adaptor conditions, for each observer and the average across observers. (C–D) The normalized results and average transfer across observers for (C) all white and (D) one-alternating polarity lines, respectively.

Figure 6. Results for Experiment 3– orientation condition.

TAEs obtained with (A) snake adaptor and test for small (light blue bars), large (dark blue bars) and no jitter (white bars) adaptor conditions, for each observer and the average across observers. (B) one-alternating orientation adaptor and test lines for small (hatched light blue bars), large (hatched dark blue bars) and no jitter (hatched white bars) adaptor conditions, for each observer and the average across observers. (C–D) The normalized results and average transfer across observers for (C) snake and (D) one-alternating orientation lines, respectively.

Figure 7. Schematic representation of a neural model the might explain the asymmetry in the TAE obtained in the polarity (A) and orientation (B) conditions, as applied to lines of 4 microelements.

The lines are all detected by 2^nd-order mechanisms that detect changes across space and/or time of the response energies of linear simple-cell-like filters (termed ‘1^st-order’) that pick up the local luminance/orientation detail in the image. The rectified 1^st-order signals are combined by the 2^nd-order filter via AND-gating operation (denoted by X). (A) The all-white and all-dark 2^nd-order responses are combined into a common pathway (denoted by blue). A second pathway is sensitive to all-white, all-black and alternating-polarity lines (denoted by red). The arrangement of pathways explains the asymmetry in the transfer of TAE between alternating-polarity adaptors combined with all-white tests, and all-white adaptors combined with alternating-polarity tests. The scheme for coding lines made from different orientations (B), is somewhat different from the scheme for coding lines with different luminance polarities in that snakes and ladders are processed by different pathways (see text for details).