Memory for Luminance: An exploration of a bias in judgements of discrete changes in luminance

Abstract

<p>The series of nine experiments in this thesis describes an exploration of what we believe to be a new empirical effect for luminance judgements. The basic effect is that judgements about discrete changes in luminance are biased in the direction of previous luminances. Along with an introductory chapter and a final chapter discussing overall conclusions, the nine experiments are arranged into three chapters, each of which addresses an hypothetical explanation of the effect.</p> <p>In Chapter 2, the relationship between the luminance bias and representational momentum (RM) was examined. Experiment 1 demonstrated the basic RM effect using a common RM paradigm (E.g. Freyd & Finke, 1984), and showed that when subjects in a comparable experiment made luminance judgements, they showed data opposite to that predicted by the RM hypothesis. Experiment 2 showed that this qualitatively different pattern of data was not the result of time course differences. Experiment 3 showed that the effect of a manipulation which reduces or eliminates RM has a vary different effect on the luminance task. We conclude from these experiments that RM may not act on the dimension of luminance change, that this finding is important in light of current debates as to the function of RM, and that the luminance effect is worthy of further study in its own right.</p> <p>In Chapter 3, we examined whether the luminance bias might be explained by either of two mechanisms which are argued to be the result of processes early in the visual system, namely the "ramp aftereffect" first reported by Antis (1967), and constancy. Because the ramp aftereffect is known not to transfer across retinal locations, Experiments 4 and 5 tested whether the luminance bias under investigation here does so. It does. Experiment 6 asked whether the size of the bias changes for dark items on a light background, as opposed to light items on a dark background, a manipulation which results in more robust lightness constancy (Jacobsen & Gilchrist, 1988b). It does not. We conclude that the luminance bias is not the result of these early perceptual mechanisms.</p> <p>In Chapter 4, we asked whether the bias for luminance judgements is better characterized as stemming from the first item in the display, the most recent item in the display, or some average of all the displays. Experiments 7 and 8 manipulate the first and most recent items while holding the average luminance of the displays constant. Experiment 9 manipulated the average luminance of the displays, while holding the most recent item constant. The results suggest that the luminance bias is affected by the average luminance of the displays. This finding puts a constraint on potential models of this luminance bias.</p> <p>The final chapter in the thesis discusses the conclusions to be drawn from the nine experiments. It argues that these findings suggest that representational momentum can only be observed for a subset of dimensions of change, contrary to current theory. It speculates on the physiological underpinnings of both RM and the luminance bias. It suggests that the luminance bias is better characterized as a memorial rather than a perceptual effect. Finally, it speculates on the possible adaptive value of this effect.</p>