Talk:Visual illusions: An Empirical Explanation
Dear Dr. Seth Dear Dr.Purves
I found the manuscript very nice, about an important issue and I enjoyed reading it.
The consequences of uncertainties in visual input, i.e. of visual feature statistics, for perception are nicely documented. However, I feel that highlighting important potentials of such an empirical approach to uncover its implementation in neuronal circuitries may even strengthen the conclusions. A few suggestions that address neurophysiological questions about the underlying functional architecture - that might be added - are given below.
Partly as the outcome of my enthusiasm, I tried to detail some facets in a longer paragraph that certainly does not meet the typical review format. I hope however that you will regard this - a bit unconventional way of statement - not as inappropriate but may find some useful aspects that could be included.
With kind regards, Dirk
The present article derives a fundamental concept of how to approach visual processing strategies in highly developed neuronal systems. It becomes clear that the term “visual illusion” is in itself misleading as it implies that the task imposed on visual sensation is to “re”-present the external world. In this view of course, any deviation from physical references must be interpreted as erroneous computing.
Instead, the task the visual system has to solve is to generate and interpret internally meaningful patterns of neuronal activity that enable successful interaction in natural environments. Visual functioning is then guided by selective organization of external input that biological systems are actually confronted with. In this framework, adaptation, stability that preserves flexibility, and prediction, are the essential capacities that have to be acquired throughout ongoing individual learning and evolutionary processes.
Besides accumulating empirical information given in environmental settings, the system foremost operates dynamically on its neuronal structure; in particular as the visual system receives persistently changing input. In that way internal dynamics must be adequately adjusted to space-time characteristics of external events.
How could visual illusions in this context be used for physiological experiments to reveal neuronal processing mechanisms and their underlying functional architecture?
It is exactly the definition of visual illusions that may be operationalized here: In a first step a given visual illusion is decomposed in terms of well-defined physical attributes. In a second step neuronal responses to the selected individual parameter settings are measured. By this means the correspondence of neuronal activity to isolated physical properties is used as the experimenters’ tool to decode particular stimulus features within neuronal responses, i.e. to “reconstruct” physical values from neuronal activity. The crucial final step is then to use deviations from a simple mapping of physical values - thus using composite illusory paradigms - in order to identify neuronal interactions (Jancke et al., 1999). Consequently, discrepancies between stimulus and its neuronal representation allow for conclusions about the underlying functional architecture of neuronal circuitries.
For example, in the flash-lag illusion (Hazelhoff & Wiersma, 1924) a single spot of light is briefly presented physically aligned with a moving light source. However, perceptually the moving stimulus appears ahead of the flash. This observation motivated physiological experiments, in which neuronal population responses to briefly presented flashes were measured to estimate representations of stimulus location in early visual cortex. By leaving the estimator fixed, activity in response to a moving stimulus revealed indeed a shift ahead of a single flash representation. Thus, a neuronal correlate of a visual illusion could be detected here (Jancke et al., 2004a).
In an empirical interpretation, any appearing light distribution may be subject to sudden onset of motion occurring at high probability due to frequent ego or object movements. The visual system should therefore have implemented various anticipatory an integrative mechanisms. For instance, it was speculated that any visual cue might provide pre-activation via long-range axonal arborization at early cortical processing stages. In fact, far reaching subthreshold propagating waves of cortical activity in response to local flashes and moving stimuli could be visualized using voltage-sensitive dye imaging (Jancke et al., 2004b), most apparent in the line-motion illusion paradigm (Hikosaka et al., 1993): Here a long line or bar, presented after a small dot, is perceived as being gradually drawn-out, most likely revealing subsequently boosting of a traveling cortical activity wave induced by the preceding dot (see figure).
Subthreshold cortical activity that spreads far ahead of the actual stimulus representation may “prepare” the cortex for an object’s putative trajectory. When actual movement occurs, neurons that represent position ahead of the stimulus are already close to firing threshold and could more rapidly react to motion. However, there is still a gap of knowledge about how timing information provided by visual cortical neurons influence perception (Seriès et al., 2002). Along visual pathways various predictive (Nijhawan, 1994; Rao & Ballard, 1999) and integrative mechanisms (Krekelberg & Lappe, 1999; Eagleman & Sejnowski, 2000) starting from the retina (Berry et al., 1999) up to sensorimotor transformation (Kerzel & Gegenfurtner, 2003) are involved in processing of motion in order to successively navigate in natural environments.
Visual motion illusions can be used to unravel such complex neuronal functional architecture that has empirically internalized typical contextual relationships of visual features in both space and time. In this sense, visual illusions provide in general experimental access in that deviations from physical features can be attributed to the underlying dynamics of neuronal mechanisms at different processing stages.
To me the list given in chapter “Approaches to understanding visual illusions” implies that these approaches could clearly be separated from an explanation in empirical terms. However, as stated, they may just not explain the “full spectrum” of visual illusions. Particularly approaches 2/3 are dependent on their interpretation and contain useful aspects of the empirical approach. This could be made clearer.
It may be mentioned that Gestalt psychologists tried to categorize general principles that could govern visual perception. Rather than a simple parameter mapping they deduced rules about coherences within natural objects. Wouldn’t this fit to the empirical approach?
Write error in last sentence ”approaches to understanding visual illusions”: should be Helmholtz, 1924
The line-motion illusion and its cortical correlate: http://www.nature.com/nature/journal/v428/n6981/extref/nature02396-s1.htm
Berry II MJ, Brivanlou, IH, Jordan TA & Meister M (1999). Anticipation of moving stimuli by the retina. Nature 398, 334-338. Eagleman DM & Sejnowski TJ (2000). Motion integration and postdiction in visual awareness. Science 287, 2036-2038. Hazelhoff FF & Wiersma H (1924). Die Wahrnehmungszeit [The sensation time]. Zeitschrift für Psychologie 96, 171-188. Hikosaka O, Miyauchi S, Shimojo S (1993). Focal visual attention produces illusory temporal order and motion sensation. Vision Res. 33, 1219–1240. Jancke D, Erlhagen W, Dinse HR, Akhavan AC, Giese M, Steinhage A & Schöner G (1999). Parametric population representation of retinal location: Neuronal interaction dynamics in cat primary visual cortex. J. Neurosci. 19, 9016-9028. Jancke D, Erlhagen W, Schöner G, Dinse HR (2004a). Shorter latencies for motion trajectories than for flashes in population responses of primary visual cortex. J Physiol 556: 971-982. Jancke D, Chavane F, Naaman S & Grinvald A (2004b). Imaging correlates of visual illusion in early visual cortex. Nature 428, 423-426. Kerzel D & Gegenfurtner KR (2003). Neuronal processing delays are compensated in the sensorimotor branch of the visual system. Curr. Biol. 13, 1975-1978. Krekelberg B & Lappe M (1999). Temporal recruitment along the trajectory of moving objects and the perception of position. Vision Res. 39, 2669-2679. Nijhawan R (1994). Motion extrapolation in catching. Nature 370, 256-257. Rao RPN & Ballard DH (1999). Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive field effects. Nat. Neurosci. 2, 79-87. Seriès P, Georges S, Lorenceau J, Frégnac Y (2002). Orientation dependent modulation of apparent speed: a model based on the dynamics of feed-forward and horizontal connectivity in V1 cortex. Vision Res. 42, 2781–2797.
I happen to be a fan of the authors’ framework; however, it seems to me that given Scholarpedia’s standards for scope (as described here: http://www.scholarpedia.org/article/Scholarpedia:Instructions_for_reviewers), this article should perhaps be narrowed to something like ‘Visual Illusions: An Empirical Explanation’, since the article is focused on that rather than the variety and history of illusions more generally. That way, an article that is more generally about visual illusions (giving examples of the giant variety of illusions, the proposed and debated explanations, and so on) could then reference this article as an explanatory framework. I recognize that the authors will argue that their framework can capture many (perhaps all) visual illusions under its umbrella – it’s simply that it may (like everything in science) turn out to be wrong, in which case it might be valuable to separate the world of visual illusions from a particular framework.
Some of the claims were too loose for a peer reviewed article. For example, one paragraph suggests that the flash-lag effect, aperture problem, and the perception of speed and direction “are likely to be explained in terms of the accumulated influence of images and their relationships to moving objects in the environment (W.Wojtach, unpublished observations)”. These are all areas with well-developed literatures. It may be the case that the authors can provide an alternative explanation – but this needs to be proven first before it can be staked as a claim in a short encyclopedic entry on visual illusions.
Finally, I found the white and black bars in Fig 4B confusing. Are these simply meant to indicate that templates were passed over all possible positions and orientations, or are they meant to indicate that these are the locations where bars (the white ones) found a good match? If the latter, it is totally unclear to the reader who has never seen the underlying photograph what elements were underneath the white bars that would provide such a match. Please clarify.