Predictive and postdictive mechanisms jointly contribute to visual awareness

Abstract

One of the fundamental issues in visual awareness is how we are able to perceive the scene in front of our eyes on time despite the delay in processing visual information. The prediction theory postulates that our visual system predicts the future to compensate for such delays. On the other hand, the postdiction theory postulates that our visual awareness is inevitably a delayed product. In the present study we used flash-lag paradigms in motion and color domains and examined how the perception of visual information at the time of flash is influenced by prior and subsequent visual events. We found that both types of event additively influence the perception of the present visual image, suggesting that our visual awareness results from joint contribution of predictive and postdictive mechanisms.

Introduction

If the world that we perceive is the result of hierarchical processing of incoming information about the world, what is perceived to be the present is actually what happened in the past. Such a world is inevitably dangerous. One way to compensate for the delay is to predict the future based on the past history of the incoming information (Cavanagh, 1997, Nijhawan, 2008, Wojtach et al., 2008). This idea is supported by the findings from a number of experiments such as omitted flash response, motion extrapolation and predictive remapping of visual features (Berry et al., 1999, Melcher, 2007, Nijhawan, 1997, Schwartz et al., 2007). However, other data suggest that the delay is not compensated: the past is merely perceived as the present (Libet, 2004). Results of experiments using paradigms of backward masking, the color phi phenomenon and the flash-lag effect in a flash-initiate condition seem to indicate the importance of a postdictive mechanism (Breitmeyer, 1984, Eagleman and Sejnowski, 2000a, Kolers and von Grunau, 1976, Nijhawan, 2002). It is unclear which of these two apparently conflicting theories, prediction and postdiction, can best explain the mechanism of how we actually perceive the world.

A flash-lag paradigm has been used to test which of these two theories is correct (Fig. 1) (Nijhawan, 2002). Typically a stimulus with a continuously changing feature such as position or color is presented (changing stimulus), with another stimulus (flash) briefly presented nearby (Fig. 1a). In this flash-lag condition (FLC) the feature value of the changing stimulus is perceived to be ahead of that of the flash, even though the feature values of both stimuli are physically identical at the time of the flash (Nijhawan, 1994, Sheth et al., 2000). This illusion in the FLC has been taken to support the predictive mechanism (Nijhawan, 1994, Nijhawan, 2008). The world that we perceive now is the one that is predicted based on the trajectory of changing features before the flash. However, it is also possible that the perception is influenced by the trajectory of the changing stimulus after the flash. The results of studies using flash-terminate condition (FTC) support the idea of postdiction. In the FTC, a flash is presented at the end of the continuously changing feature of another stimulus (Fig. 1b). It has been shown that subjects accurately perceive the feature of the changing stimulus (Nijhawan, 2002, Nijhawan, 2008). The results obtained from this condition suggest that events preceding the flash do not influence the perception of the changing stimulus- rather it is the events after the flash that affect perception. This idea of postdiction is further supported by the result of flash-reverse condition (FRC), where the trajectory of the continuously changing feature suddenly reverses at the time of the flash (Fig. 1c). In the FRC, the changing stimulus is perceived to be deviated toward the feature after the flash. The flash-lag illusion is also observed in a flash-initiate condition (FIC), where the flash appears at the onset of the continuous changing feature, that is, when there is no preceding event before the flash (Fig. 1d) (Eagleman and Sejnowski, 2000a, Nijhawan, 2002). These findings have been taken to support the postdictive mechanism, and reject the contribution of the predictive mechanism to visual awareness (Eagleman & Sejnowski, 2000a). Thus, in most of the previous studies prediction and postdiction have been treated as mutually exclusive ideas. Some have mentioned the possibility of a complementary role between the two mechanisms (Eagleman, 2008, Grush, 2005, Maus and Nijhawan, in press), but the idea has not been tested experimentally. Moreover it remains open whether or not the two mechanisms operate in an additive and independent manner.

In the present study we show that the two mechanisms coexist and contribute additively to perception. First, contrary to the previous studies, we show that the flash-lag illusion is observed even in the FTC especially when the target stimulus is ambiguous. This finding was confirmed in both the motion and color domains, suggesting that the predictive mechanism contributes to perception regardless of the stimulus domain. We also show that, by comparing the magnitudes of the flash-lag illusion in the FLC, FTC, FRC and FIC, the events that precede the flash and the events that follow the flash additively affect perception of a stimulus with changing features. Finally we show that the speed of the changes in the stimulus feature differently affect the magnitude of the flash-lag illusion between the FTC and FIC, at least in the color domain. This suggests a possibility that the prediction and postdiction are subserved by independent mechanisms.