Stimulus-contingent analysis produced positive amplitude values (see
Figures 4 and
5), consistent with assimilative effects; however, the values for high noise and high noise to low were nonsignificant (Low noise to low noise,
a = 2.14°,
p = 0.001, Root Mean Squared Error = 0.82; High noise to high noise,
a = 2.25 degrees,
p = 0.098, RMSE = 1.27; High noise to Low,
a = 0.98 degrees,
p = 0.059, RMSE = 0.54; Low noise to High,
a = 2.4 degrees,
p = 0.006, RMSE = 0.7). As a form of control analysis, the DoG fitting procedure for stimulus-based analysis was applied to participant errors conditioned on the relationship between orientation in current and future trials (t+1 analysis). No meaningful relationship should be observed, barring precognition. As expected, this analysis produced no significant results (all DoG amplitudes, in all conditions,
p > 0.10).
Similarly, response-contingent analysis also yielded positive DoG amplitude values, however in this case all were significant (Low noise to low noise, a = 1.68 degrees, p = 0.003, RMSE = 0.56; High noise to high noise, a = 3.44 degrees, p < 0.001, RMSE = 0.77; High noise to Low, a = 1.7 degrees, p = 0.003, RMSE = 0.43; Low noise to High, a = 2.59 degrees, p = 0.005, RMSE = 0.78).