Results

Mean percentage of time perceiving coherent motion under the 2 (attention) × 3 (monocular depth) conditions is plotted in Figure 17a. Two-way ANOVA shows that the main effect of attention is significant (F(1,10) = 11.44, p < 0.01, partial η² = 0.534). The percentage of time perceiving coherent motion is higher in the attending-to-occluders condition (76.36%) than in the attending-to-moving lines condition (60.62%), which is consistent with the hypothesis. The effect of monocular depth is also significant (F(2,20) = 65.62, p < 0.001, partial η² = 0.868). This is also consistent with the prediction, indicating that the manipulation of monocular depth is valid. The interaction of the two factors is also significant (F(2,20) = 7.82, p < 0.01, partial η² = 0.439). However, the simple main effects show that the effect of attention is not only significant at the normal condition (F(1,10) = 9.62, p < 0.05) but also significant at the blurred-occluders condition (F(1,10) = 10.58, p < 0.01) and marginally significant at the blurred-moving-lines condition (F(1,10) = 4.20, p = 0.068). These results conflict with the prediction that the effect of attention should disappear under the blurred-moving-lines and blurred-occluders conditions.

The two-way ANOVA of RT to the probe in eight blocks shows that the interaction is not significant as predicted (F(1,10) = 0.24, p = 0.637), implying that

‧

participants did not fully follow the instructions to allocate their attention. Therefore, only the data in which the participants followed the instructions to allocate their attention were selected for further analysis.

The selection method was the same as in the Experiment 2. The eight blocks were divided into four sections (Section 1: Blocks 1 and 2; Section 2: Blocks 3 and 4, and so on). Each section contained an attending-to-occluders block and an attending-to-moving-lines block to check whether the index of attention (as described in Experiment 1b) in each section is positive. Only the sections in which the index of attention is positive—implying that participants allocated their attention according to the instructions—are selected for further analysis.

There were 21 selected sections (total: 4 section × 11 participants = 44 section;

the data of one participant was not all selected), which contained 11 attending-to-occluders-first and attending-to-moving-lines-next sections and 10 attending-to-moving-lines-first and attending-to-occluders-next sections. Thus, the confounding of the order effect would be very small. The mean percentage of time perceiving coherent motion under the 2 (attention) × 3 (monocular depth) conditions is plotted in Figure 17b. Two-way ANOVA shows that the main effect of attention is significant (F(1,9) = 8.42, p < 0.05, partial η² = 0.483). The percentage of time perceiving coherent motion is higher in the attending-to-occluders condition (76.52%)

‧

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

than the attending-to-moving-lines condition (61.90%), which is consistent with the hypothesis. The effect of monocular depth is also significant (F(2,18) = 40.50, p <

0.001, partial η² = 0.818). Tukey’s HSD post hoc test (q.95(3,18) = 3.61, HSD = 21.39) shows that all paired comparisons are significant. The percentage of time perceiving coherent motion is highest in the blurred-moving-lines condition (96.44%) and lowest in the blurred-occluders condition (43.15%), with the value of the normal condition between the two (68.04%). This is also consistent with the prediction, indicating that the manipulation of monocular depth is valid.

The most important aspect is that the interaction of the two factors is significant (F(2,18) = 3.96, p < 0.05, partial η² = 0.306). The simple main effects show that the effect of attention is only significant at normal conditions (F(1,9) = 18.22, p < 0.01) and not significant at the blurred-occluders condition (F(1,9) = 2.23, p = 0.169) or blurred-moving-lines condition (F(1,9) = 0.82, p = 0.390). These results are consistent with the prediction that the effect of attention should disappear under the blurred-moving-lines and blurred-occluders conditions and that the effect of attention should remain constant under normal conditions.

‧

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

(a)

(b)

Figure 17. Results of Experiment 4. (a) Means and 1 standard errors of the percentage

‧

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

(attention) × 3 (monocular depth) conditions in eight blocks. (b) Means and 1 standard errors of the percentage of time that the diamond stimulus was perceived as coherent motion under the 2 (attention) × 3 (monocular depth) conditions in the selected sections.

‧

The results show that the monocular depth cue (by filtering out HSF information) can eliminate the effect of attention, which is consistent with the prediction in the hypothesis that attention can bias multistable motion perception through affecting the depth perception mechanism. By filtering out the HSF information of the four moving lines, the effect of attention almost completely disappears under the blurred-moving-lines condition (the difference of the percentage of time perceiving coherent motion under the two attention conditions is smaller than 0.70%). This implies that it is difficult for attention to influence multistable motion perception through affecting the depth perception mechanism because the monocular depth cue already defines the depth relationship of the occluders and lines. Since most participants indicated that they can attend to demanded areas even if they look blurred, the reason for the effect of attention disappearing under the blurred-moving-lines condition cannot be explained by participants being unable to attend to them.

Although there is still some effect of attention under the blurred-occluders condition (the difference of the percentage of time perceiving coherent motion under the two attention conditions is about 16.55%), it is not significant (power = 0.720, which is estimated by Cohen’s (1988) medium effect size f = 0.25). Also, the effect of

‧

attention is much smaller than under normal conditions (26.59%). This implies that the effect of attention is weakened (but not completely blocked) due to the monocular depth cue. One reason might be that even in the blurred-occluders condition, there were still some monocular depth cues (such as T-junctions) and structures of the diamond stimulus to let the moving lines be perceived as completed behind the occluders, so the occluders cannot be seen as entirely behind the moving lines as expected.

An interesting question is that why the effect of attention showed in the blurred-occluders condition and blurred-moving-lines condition before deleting the negative “index of attention” sections, which should imply that participants did not allocate their attention on demanded areas. One reason may be that participants only attended to one moving line among the four. Accordingly, the “index of attention” in these sections became negative because the probe might show on other lines that they did not attend. This may cause the percentage of time perceiving coherent motion to decrease under the attending-to-lines condition because it is hard to perceive the four lines connecting to each other and moving coherently if only one line is attended. And this decrease may be the reason why the effect of attention showed in the blurred-occluders condition and blurred-moving-lines condition before deleting the negative “index of attention” sections.

‧

In the multistable perception domain, many studies have found that directing fixation or spatial attention to different regions can bias the perception of the Necker cube (Toppino, 2003; Kawabata, 1986, 1987; Meng & Tong, 2004; Xu & Franconeri, 2010). The explanation in these studies is that the fixated or attended region would look closer, so the perception is biased. However, no study has examined this assumption directly. On the other hand, in research concerned with attention, few studies have mentioned whether spatial attention can affect depth perception. Some studies have found that spatial attention can influence many properties of stimulus perception, which may relate to depth perception (e.g., Carrasco, 2006; Carrasco, Williams, & Yeshurun, 2002; Gobell & Carrasco, 2005). Thus, it is possible for spatial attention to affect multistable perception through the depth mechanism.

The purpose of this study is to investigate the effect of spatial attention on multistable figure perception and the possible underlying mechanism of this effect.

The diamond stimulus was used to test the hypothesis that spatial attention can bias multistable motion perception by making attended areas look closer. Four experiments are designed to examine this hypothesis and the main results will be briefly reviewed first.