4 The electrophysiological correlate of contour integration is modulated by task
5.4 Results
5.4.2 ERP- measurements
Studie 2: ERPs for Colour and Luminance
Time-window analysis: The details of the significant results of the time-window analysis for experiment 1 are given in Tab. 5-1. With the time window starting 230 ms after stimulus on-set detectable FIGUREs elicited a larger negative shift than CONTROL displays. This nega-tive shift still occurred at the end of the investigated time window (350 ms after stimulus on-set). This was indicated by a significant effect of level (F(1,7) = 12.2-54.1, p<.015), see Fig.
5-6 and Fig. 5-7. Furthermore, the contour-specific negative response was larger for the CIR-CLE than for open contour types. This was indicated by significant interactions between con-tour types x level between 280 and 300 ms (F(3,21) = 6.2-6.3, p<.015), see Fig. 5-8, left. No differences were found regarding the scalp distribution.
Experiment 1 200- 210- 220- 230- 240- 250- 260- 270- 280- 290- 300- 310- 320- 330- 340- 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 ms
contrast
df (1.7)
level 12.2 14.0 17.0 19.8 26.5 36.1 37.7 43.7 50.5 54.1 45.5 32.8
df (1.7) 0.010 0.007 0.004 0.003 0.001 0.001 0.000 0.000 0.000 0.000 0.000 0.001
contrast x level
df (1.7)
contour type
df(3.21)
contour type x level 6.2 6.3
df (3.21) 0.010 0.009
Tab. 5-1 indicates the significant effects of the time-window analysis for experiment 1. The time-windows inves-tigated are depicted from left to right; the corresponding time-periods of each window are stated in the top row.
Stimulus onset correspond to 0 ms. The effects investigated and corresponding degrees of freedom (df …) are stated in the left column. Within each cell significant results are indicated by grey shading. The F-values are depicted on the top and the corresponding p-values on the bottom of each cell indicating significant effects.
Empty cells indicate lack of significant effects.
Fig. 5-6 Grand average ERPs elicited by the FIGURE and CONTROL task conditions of the RED/GREEN run during experiment 1. Zero marks the stimulus onset. Over posterior recording sites the ERP waveforms are shifted more towards negative amplitude values for the FIGURE than for the CONTROL task condition.
Fig. 5-7 Grand average ERPs elicited by the FIGURE and CONTROL task conditions of the LUMINANCE run
Studie 2: ERPs for Colour and Luminance
Fig. 5-8 Grand average ERPs elicited by the CIRCLE (top row) and CONCAVE contours (bottom row) for experiment 1 (left) and group 1 of experiment 2 (right). The grand averages are shown for the electrode site PO7. The two LEVELs (FIGURE and CONTROL) are superimposed in each plot. Zero marks the stimulus onset.
Summary of experiment 1
During the EEG-experiment, the N75 was elicited only for the LUMINANCE displays. The contour-specific response started approximately 230 ms after stimulus onset as a negative shift elicited by detectable contours. This shift was largest for the CIRCLE. No differences between LUMINANCE and RED/GREEN were found with respect to contour detection (see Fig. 5-9). Experiment 2 was conducted to investigate if this lack of modulation was due to remaining luminance cues hidden in the RED/GREEN displays. Experiment 2 masked and hence eliminated any remaining luminance artefacts for RED/GREEN displays on which con-tour integration mechanisms may have relied on in experiment 1.
Fig. 5-9 depicts representative averages of three individuals for the electrode site PO7; one individual for each experimental group (experiment 1 – left, group 1 of experiment 2 – middle and group 2 of experiment 2 – right).
In the top row the FIGURE ERPs for RED/GREEN and LUMINANCE are shown. In the middle row the URE ERP for RED/GREEN is shown together with its CONTROL ERP. Similarly, in the bottom row the FIG-URE ERP for LUMINANCE is now shown together with its CONTROL ERP.
All three subjects elicit a contour-specific negative shift for colour- and luminance-defined contours (middle and bottom row). In the time-window where the contour-specific negative response is elicited the FIGURE ERPs are highly similar for RED/GREEN and LUMINANCE ERPs (indicated by arrows for experiment 1). For group 2 of experiment 2 the ERP for RED/GREEN was shifted more towards negative values than for LUMINANCE.
EEG-experiment of experiment 2
Early sensory VEP: As in experiment 1, the N75-amplitude for LUMINANCE was larger than for RED/GREEN as indicated by a significant effect of contrast (F(1,14) = 16.3, p<.01), see Fig. 5-5, right. Furthermore, LUMINANCE elicited a higher positive P1 amplitude than RED/GREEN as was indicated by a significant effect for contrast for the P1 amplitude (F(1,14) = 33.6, p<.001).
Time-window analysis: The time-window analysis in experiment 1 yielded a significant modulation of the ERP by contour integration, i.e. by the FIGURE task condition. This effect might be due to the saliency of the contours(Mathes et al. 2006b). In experiment 2 the
Studie 2: ERPs for Colour and Luminance
physical results). We, therefore, conducted the time-window analysis for experiment 2 sepa-rately for groups 1 and 2.
Time-window analysis – group 1: The details of the significant results of the time-window analysis for group 1 of experiment 2 are given in Tab. 5-2. The contour-specific negative shift started 220 ms and was still observed 340 ms after stimulus onset (F(1,7) = 14.7-36.0, p<.01), see Fig. 5-10 and Fig. 5-11. Closed contours, especially CIRCLEs, elicited more negative amplitudes than open contour types between 230 and 250 ms after stimulus onset as was indi-cated by a significant effect for contour type (F(3,21) = 5.8-6.9, p<.01), see Fig. 5-8, right.
Similarly, the significant interaction between contour type x level occurring between 270 and 280 ms after stimulus onset indicated that the contour-specific negative shift showed its high-est prominence for closed compared to open contour types (F(3,21) = 6.3, p<.01). No differ-ence was found for contour detection between LUMINANCE and RED/GREEN (see Fig.
5-9).
Experiment 2 200- 210- 220- 230- 240- 250- 260- 270- 280- 290- 300- 310- 320- 330- 340- Group 1 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 ms
contrast
df (1.7)
level 15.5 18.6 16.1 25.8 30.2 26.6 36.0 17.8 14.7 16.9 16.2 16.7 df (1.7) 0.006 0.004 0.005 0.001 0.001 0.001 0.001 0.004 0.006 0.005 0.005 0.005 contrast x level
df (1.7)
contour type 5.8 6.9
df(3.21) 0.009 0.008
contour type x level 6.3
df (3.21) 0.007
Tab. 5-2 indicates the significant effects of the time-window analysis for group 1 of experiment 2. See legend of table 1 for further information.
Fig. 5-10 Grand average ERPs elicited by the FIGURE and CONTROL task conditions of the RED/GREEN run for group 1 of experiment 2. Zero marks the stimulus onset.
Studie 2: ERPs for Colour and Luminance
Time-window analysis – group 2: The details of the significant results of the time-window analysis for group 2 of experiment 2 are given in Tab. 5-3. The ERP for RED/GREEN was more negative than for LUMINANCE between 290 and 310 ms after stimulus onset. This was indicated by a significant effect for contrast (F(1,7) = 11.5-12.0, p<.015), see Fig. 5-9.
For group 2 of experiment 2 the negative shift for detectable contours started 250 ms after stimulus onset and was still observed 350 ms after stimulus onset, as was indicated by a sig-nificant effect for level (F(1,7) = 14.7-28.1, p<.01), see Fig. 5-12 and Fig. 5-13. Furthermore, the analysis of the scalp distribution revealed a more negative amplitude over occipital elec-trode sites for LUMINANCE than for the RED/GREEN stimuli during the time window be-tween 200 and 210 ms after stimulus onset. This was indicated by a significant contrast x electrode interaction (F(10,70) = 6.4, p<.01).
Experiment 2 200- 210- 220- 230- 240- 250- 260- 270- 280- 290- 300- 310- 320- 330- 340- Group 2 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 ms
contrast 11.5 12.0
df (1.7) 0.011 0.010
level 28.1 27.3 14.7 19.5 24.3 19.3 19.2 16.0 15.7 18.4
df (1.7) 0.001 0.001 0.006 0.003 0.002 0.003 0.003 0.005 0.005 0.004
contrast x level
df (1.7)
contour type
df(3.21)
contour type x level
df (3.21)
Tab. 5-3 indicates the significant effects of the time-window analysis for group 2 of experiment 2. See legend of table 1 for further information.
Fig. 5-12 Grand average ERPs elicited by the FIGURE and CONTROL task conditions of the RED/GREEN run for group 2 of experiment 2. Zero marks the stimulus onset.
Studie 2: ERPs for Colour and Luminance
LUMINANCE-CONTROL: This analysis was conducted to examine whether possible re-maining luminance artefacts in the RED/GREEN stimuli may explain the high similarity of the contour-specific negative shift in both the LUMINANCE and RED/GREEN conditions.
All experiments elicited a negative shift for detectable FIGUREs in overlapping time win-dows. Therefore, we combined the two datasets of both groups of experiment 2 in order to increase the power of analysis for the LUMINANCE-CONTROL task condition. No signifi-cant modulations (main and interaction effects) were found between LEVELs or CONTOURs (see Fig. 5-14).
Fig. 5-14 Grand average ERPs elicited by the FIGURE and CONTROL task conditions of the LUMINANCE-CONTROL run during experiment 2. Zero marks the stimulus onset. ERPs of the FIGURE and LUMINANCE-CONTROL task conditions are highly similar.
Summary of experiment 2
Luminance-defined stimuli elicited more prominent components of the early ERP, i.e. the N75 was elicited only for the LUMINANCE stimuli and P100 was larger for LUMINANCE than for RED/GREEN stimuli.
hancement resulted from a higher saliency for the closed contours, which were hidden since all participants of group 1 performed near ceiling levels. For group 2 of experiment 2, we therefore optimised our procedure to individually adjust performance between contour types and allowed the orientation jitter α to exceed 10°, unlike in the other experimental groups. For group 2 the contour-specific negative shift occurred first 250 ms after stimulus onset, i.e. in a later time window than for group 1, and was not modulated by contour type.
As in experiment 1 the electrophysiological correlate of contour integration was similar for LUMINANCE and RED/GREEN in both groups. Although for group 2 the ERP of the RED/GREEN stimuli was more negative in the time window during which the specific negative response occurred, we found no qualitative difference between the contour-specific response for LUMINANCE and RED/GREEN (see Fig. 5-9, Fig. 5-12 and Fig. 5-13).
A great advantage of the luminance mask was that it allowed us to directly measure the effec-tiveness in masking luminance-defined contours. A control experiment demonstrated that any remaining luminance cues in the RED/GREEN condition were masked up to at least a Michelson contrast of 10 %. The photometrically measured luminance contrast of the RED/GREEN stimuli was below a Michelson contrast of 0.01 %. Any possibly remaining luminance signal in RED/GREEN stimuli was therefore sufficiently masked by the luminance noise superimposed on all displays during experiment 2. Contour integration in RED/GREEN stimuli had to exclusively rely on orientation cues derived from the colour contrasts of the Gabor elements.