Dear Editor,
Kagan's recent response to our article lists a number of alleged "methodological and conceptual issues". Here we show that Kagan's concerns arise from problematic logic, conceptual misunderstandings, and inconsistent analyses of the literature, including his own published research, and we take the opportunity to clarify these sources of confusion.
1) Kagan claims that our study does not...
Dear Editor,
Kagan's recent response to our article lists a number of alleged "methodological and conceptual issues". Here we show that Kagan's concerns arise from problematic logic, conceptual misunderstandings, and inconsistent analyses of the literature, including his own published research, and we take the opportunity to clarify these sources of confusion.
1) Kagan claims that our study does not address "restoration of visibility" or "fading", but perceptual contrast sensitivity.
This statement is factually incorrect. First, in contrast sensitivity experiments, subjects are tasked with detecting the presence of visual stimuli with physically varying contrasts (Felipeet al., 1993; Li et al., 2009), whereas participants in our study reported on the illusory fading of a physically invariant stimulus (a perceptual phenomenon known as "Troxler fading" since Troxler's 1804 description). We note that Troxler fading, the perceptual experience at the heart of our study, is a gradual, rather than an instantaneous process. Often, an object becomes less and less visible until it eventually disappears (and then reappears, typically when microsaccades bring it back). Other times, an object's visibility decreases at first, and then it is restored (again, usually in connection with microsaccade production) before complete fading has occurred. One of the main strengths of our study is that it quantifies the precise timing of the interactions between ocular events and perceptual experience. Had we considered only "complete fading" and ignored "partial fading" events, we would have achieved an incomplete picture of the role of microsaccades in visual restoration, rather than the full, dynamic picture of the interactions between microsaccade production and the ongoing perceptual experience that characterizes natural vision. Thus, our statement that our study is concerned with perceptual fading (and the subsequent restoration of visibility) is accurate, both for experimental and historical reasons.
2) Kagan claims that that we designed our experimental conditions to induce as much fading as possible.
This is also incorrect. The stimulus was as in Martinez-Conde et al. (2006) where "The target's parameters were chosen so that it tended to either fade or intensify with an approximate 50-50 ratio during fixation." Indeed, Figure 1B in the current study shows similar distributions of fading and intensifying periods, as does Figure 1C in the 2006 study. Thus, experimental conditions were not designed to induce "as much fading as possible", but to produce a comparable amount of data in fading and intensification conditions (and thus achieve a balanced experimental design).
3) Kagan claims that our study does not apply to "real" microsaccades, as they are normally defined in the literature.
The issue of what constitutes an appropriate definition of microsaccade is comprehensively discussed in Martinez-Conde et al. (2013). We reprint some of that discussion here (please see the original article for more extensive details and references):
"Until the 1990s, microsaccades were defined as having amplitudes smaller than 12 arc min. This cut-off value originated in earlier studies finding that the distribution of saccadic sizes during fixation declined sharply around 12 arc min. However, later studies found that microsaccade sizes frequently exceed this value (instead, current microsaccade magnitude distributions often asymptote around 1 degree). Thus, most contemporary researchers have adopted the convention of using a 1-degree upper magnitude threshold (which captures more than 90% of saccades produced during attempted fixation). [...] Whatever the explanation behind the recent shift towards larger microsaccades, the combined evidence indicates that setting a cut-off value of 12 arc min is both conceptually and experimentally unjustified."
Table 1 of Martinez-Condeet al. (2009), moreover shows that, of the +30 original research studies on microsaccades conducted from 2004 to 2009, none used a maximum amplitude threshold lower than 1 degree. Thus, our use of a 1-degree threshold is in full alignment with the standards of contemporary microsaccade research. Indeed, Kagan himself has used a much less stringent 2.4 degree threshold to study "fixational saccades" (another name for microsaccades) (Snodderly et al. 2001). Thus, Kagan's present definition of what constitutes a valid microsaccade is at odds with his own research.
4) Kagan states that only 23% of <1 deg saccades preceded increasing visibility reports.
This is factually incorrect again. Our study does not report such a statistic, and we do not know how Kagan obtained this number. Perhaps he did not realize that, as explained in our paper, "microsaccadic events" may contain more than one microsaccade (indeed, paired microsaccades were more likely to occur in the peak interval than in the control region). Whatever the case may be, we have now done this calculation (not included in our original report) and found that, on average, 50% +/- 10% SEM of all microsaccades produced during the eligible region (i.e. during the fading period, where microsaccades are able to restore faded vision) occurred during the peak interval (i.e. preceding increased visibility reports).
5) Kagan claims that the fovea evolved to detect fine, high-contrast features, whereas our visual target was "large and fuzzy".
There are at least three problems with Kagan's reasoning. First, our 40% contrast Gabor patch, whether "large and fuzzy" or not, was a very visible stimulus (i.e. nowhere near perceptual or physiological thresholds) that fell on the fovea and stimulated its receptors -- until neural adaptation set in, leading to visual fading. (Incidentally, the discovery of foveal fading is several hundred years old and has been replicated numerous times (Darwin, 1795; Krauskopf, 1963; Pessoa and De Weerd, 2003; Simons et al., 2006; Troxler, 1804)). Second, the central region of the Gabor, where it visibly transitioned from dark to light, was centered over the fovea. Third, Kagan's assumption that only high-contrast stimuli are pertinent to foveal vision is dubious. On the contrary, we posit that most visible stimuli are relevant to vision, by definition. High contrast or not, fading of any stimuli is a visual degradation (which microsaccades often supersede), and there are many low contrast objects that one would want to see with central/foveal vision. Trying to find a diamond earring on a white carpet, or inspecting small features in medical and research images, are two of many examples. Thus, the fovea's ability to inspect fine features is not in conflict with its capacity to see stimuli of all contrasts and spatial frequencies.
6) Kagan says that we incorrectly state that our current findings about microsaccade directions are consistent with our previous neurophysiological data.
Here Kagan consults, and misinterprets, the wrong reference. The present study reports that microsaccades of all directions were equally effective, "in agreement with previous physiological analyses of microsaccade direction on the firing of V1 neurons (Martinez-Conde et al., 2000)". Instead of reading (Martinez-Conde et al., 2000), as cited, Kagan consulted (Martinez-Conde et al., 2002), which studied the neural responses to microsaccades in the presence of optimally and non-optimally oriented targets, but did not investigate microsaccade direction in any sense or form. Further, Kagan seemingly confuses the concepts of "microsaccade direction" and "visual target orientation" in his analysis of our 2002 paper. Had he consulted the 2000 paper as indicated, he would have found the Results section to report "no clear relationship between microsaccade direction and response magnitude". Thus, our statement, that the current psychophysical results match the previous neurophysiological data, is correct.
7) Kagan claims that subjects fixated for 30s in our study, whereas natural fixation durations are ~300 ms. He further states that fixations of 300 ms are too short for fading to take place, and that fading reports in our study occurred only every 10-13 s.
Kagan's criticism is again inconsistent with his own research, in addition to containing factually incorrect statements and flawed logic. First, depending on the nature of the task at hand, fixation durations can span a second or longer (Henderson and Hollingworth, 1999; Henderson and Pierce, 2008; Henderson and Smith, 2009; Otero-Millan et al., 2008). Second, fixation durations as reported in the literature depend critically on how one defines "fixation". That is, if one considers that a microsaccade breaks fixation (as numerous studies do), then fixations will rarely exceed 1 second. By the same token, though subjects in our study attempted to fixate for 30s at a time, strict fixations were seldom longer than 1s, given that subjects continued to produce microsaccades every few hundred ms during their fixation attempts. Third, it is easy to think of natural vision tasks that may require the observer to maintain fixation on a small target for 30 s and longer: attempting to remove a small splinter from one's fingertip with a pair of tweezers is one such example. Fourth, having subjects fixate for multiple seconds at a time is a standard experimental design feature in the fixational eye movement literature, including in Kagan's own research, where monkeys fixated for 5 seconds (Kagan et al., 2008; Snodderly et al., 2001): that is, 17 times the average fixation duration in natural vision, according to Kagan's present assertion. In our study, subjects attempted to fixate for multiple seconds, so as to produce sufficient data in a controlled experimental setting without the influence of potentially confounding factors -- just as subjects in Kagan's published research did. Fifth, Kagan neglects the key issue that visual fading in human vision has been known to happen for over 200 years (Darwin, 1795; Troxler, 1804; Pessoa & De Weerd, 2003), in natural viewing conditions -- even in the presence of fixational eye movements, head movements, and incidental body movements. Thus, fading is an important problem that the visual system must overcome in everyday vision. Sixth, the frequency of fading reports in our study was inversely proportional to the duration of the fading percept: that is, one main reason that fading reports occurred every 10s on average, as Kagan indicates, is that the stimulus was consequently faded for an average 9.5 +/- 1.8 sec out of every 30 s trial. Seventh, Figure 1B shows that 29% of all the intensification periods had durations of less than a second; that is, the target faded in less than a second 29% of the time, thus invalidating Kagan's concerns about significance to fading in natural vision.
8) Kagan claims that our study may have interpreted corrective saccades as preventing fading, yet visual fading does not trigger microsaccades.
Here Kagan again confuses dissimilar concepts and misinterprets our results. First, our study neither analyzed, nor was it concerned with, the role of microsaccades on the prevention of fading. We investigated solely the ability of microsaccades (and other eye movements) to counteract (i.e. reverse) fading, that is, to restore the visibility of already faded objects. Counteracting and preventing fading are two different issues: indeed, we are currently conducting research that addresses, for the first time, how fixational eye movements work to prevent (rather than counteract) fading. Our results to date indicate that, whereas microsaccades counteract fading once it has occurred, both microsaccades and drift work together to prevent fading before it happens. That is, fixational eye movements serve to prevent fading, but not perfectly. Microsaccades have the ability to bring the stimulus back to perception, when fading does occur. Second, we do not claim, as Kagan suggests, that all microsaccades restore fading. As previously stated, we found that ~ 50% of all microsaccades produced in the period of time in which they could have reversed fading, did so. How many of those microsaccades served to correct fixation errors, in addition, is beside the point (and was not studied in our paper). As discussed in our study, as well as in our recent Nature Reviews Neuroscience article, microsaccades may serve multiple, non -exclusive functions during fixation, just as saccades do during exploration. Finally, microsaccades do not need to be "triggered by fading" in order to restore a faded percept, and nowhere do we claim that fading triggers microsaccades: we simply show that microsaccades restore the visibility of faded percepts.
9) Kagan claims that the analogy between microsaccades/visibility and retina/LGN firing is misleading, and also that the finding that induced image fading does not increase microsaccade production (Poletti & Rucci, 2010) means that microsaccades do not counteract fading.
This reasoning is fraught with logical errors and inconsistencies. First, our analogy exemplifies how flawed logic can lead to important errors in the interpretation of research results, as explained in our study (and also below). Second, Poletti and Rucci's observations that simulated fading (artificially accomplished by physical contrast decrements and retinal stabilization techniques) does not raise microsaccade rates, have no bearing on our findings that microsaccades counteract Troxler fading (i.e. the spontaneous, illusory fading that occurs in natural vision during fixation, as described by Troxler +200 years ago). Their results merely decrease the likelihood of a mechanism that detects fading and triggers a microsaccade in response. Thus, Poletti and Rucci's results do not rule out that one function of microsaccades is to counteract fading, as discussed in our paper. We also point out that, in Poletti and Rucci's study, microsaccades still occurred while the stimulus faded artificially by physical means or was stabilized on the retina. The fact that microsaccades did not restore the visibility of such a stimulus, when turned off physically, is hardly surprising, however, and has little connection to our experimental conditions or to natural vision. Finally, we must draw attention to the fact that Poletti and Rucci' subjects fixated their gaze for 10 seconds with one eye only (the other eye was patched) and their teeth clamped around a bite bar, whereas subjects in our study used a simple forehead/chin rest during the binocular viewing of a physically unchanged stimulus. Thus, Kagan's claims that our experimental conditions do not mimic natural vision are specious, given that the research he uses to support his argument was conducted in much more artificial conditions than those in our study.
To sum up, Troxler fading has been known to occur --outside of the laboratory-- for over 200 years (Troxler, 1804; Pessoa & De Weerd, 2003). Recent experiments have demonstrated that fading does indeed occur in unconstrained viewing conditions (i.e. without the use of any means of head fixation), and that microsaccades restore visibility in that situation as well (Martinez-Conde et al., 2006). Our present study extends these findings to show that microsaccades have the ability to counteract both foveal and peripheral fading. We have moreover quantified, with a principled approach, the efficacy and contribution of microsaccades to restoring faded vision. Here we have shown that Kagan's claims to the contrary arise from problematic logic, data misinterpretations, and inconsistent analyses of the literature, including his own research. One final point to elucidate is that, contrary to Kagan's assertion, our study does not underestimate the importance of drift, but it provides the first principled analysis of its contribution to restoring faded vision. While our data do show that drift does not contribute strongly to reversing fading, its ability to prevent fading may be another issue altogether. Indeed, our ongoing research indicates that both drift and microsaccades combine to prevent fading from occurring most of the time. Please stay tuned for our upcoming paper on this research.
References
Darwin, E. (1795). Zoonomia; or the Laws of organic life, in three parts, Vol 2 (4th American ed.). Philadelphia, PA, US: Edward Earle.
Felipe, A., Buades, M. J., & Artigas, J. (1993). Influence of the contrast sensitivity function on the reaction time. Vision Research, 33(17), 2461-2466. doi:10.1016/0042-6989(93)90126-H
Henderson, J M, & Hollingworth, A. (1999). High-level scene perception. Annual Review of Psychology, 50, 243-271. doi:10.1146/annurev.psych.50.1.243
Henderson, John M., & Pierce, G. L. (2008). Eye movements during scene viewing: Evidence for mixed control of fixation durations. Psychonomic Bulletin & Review, 15(3), 566-573. doi:10.3758/PBR.15.3.566
Henderson, John M., & Smith, T. J. (2009). How are eye fixation durations controlled during scene viewing? Further evidence from a scene onset delay paradigm. Visual Cognition, 17(6-7), 1055-1082. doi:10.1080/13506280802685552
Kagan, I., Gur, M., & Snodderly, D. M. (2008). Saccades and drifts differentially modulate neuronal activity in V1: Effects of retinal image motion, position, and extraretinal influences. Journal of Vision, 8(14). doi:10.1167/8.14.19
Krauskopf, J. (1963). Effect of retinal image stabilization on the appearance of heterochromatic targets. Journal of the Optical Society of America, 53(6), 741. doi:10.1364/JOSA.53.000741
Li, R., Polat, U., Makous, W., & Bavelier, D. (2009). Enhancing the contrast sensitivity function through action video game training. Nature Neuroscience, 12(5), 549-551. doi:10.1038/nn.2296
Martinez-Conde, S., Macknik, S. L., & Hubel, D. H. (2000). Microsaccadic eye movements and firing of single cells in the striate cortex of macaque monkeys. Nature Neuroscience, 3(3), 251-258.
Martinez-Conde, S., Macknik, S. L., & Hubel, D. H. (2002). The function of bursts of spikes during visual fixation in the awake primate lateral geniculate nucleus and primary visual cortex. Proceedings of the National Academy of Sciences, 99(21), 13920-13925.
Martinez-Conde, S., Macknik, S. L., Troncoso, X. G., & Dyar, T. A. (2006). Microsaccades counteract visual fading during fixation. Neuron, 49(2), 297-305. doi:10.1016/j.neuron.2005.11.033
Martinez-Conde, S., Macknik, S. L., Troncoso, X. G., & Hubel, D. H. (2009). Microsaccades: a neurophysiological analysis. Trends in Neurosciences, 32(9), 463-475. doi:10.1016/j.tins.2009.05.006
Martinez-Conde, S., Otero-Millan, J., & Macknik, S. L. (2013). The impact of microsaccades on vision: towards a unified theory of saccadic function. Nature Reviews Neuroscience, 14(2), 83-96. doi:10.1038/nrn3405
Otero-Millan, J., Troncoso, X. G., Macknik, S. L., Serrano-Pedraza, I., & Martinez-Conde, S. (2008). Saccades and microsaccades during visual fixation, exploration and search: foundations for a common saccadic generator. Journal of Vision, 8, 14-21.
Pessoa, L., & De Weerd, P. (2003). Filling-in: From perceptual completion to cortical reorganization. New York, NY, USA: Oxford University Press.
Poletti, M., & Rucci, M. (2010). Eye movements under various conditions of image fading. Journal of Vision, 10(3), 1-18. doi:10.1167/10.3.6
Simons, D., Lleras, A., Martinez-Conde, S., Slichter, D., Caddigan, E., & Nevarez, G. (2006). Induced visual fading of complex images. Journal of Vision, 6(10). doi:10.1167/6.10.9
Snodderly, D. M., Kagan, I., & Gur, M. (2001). Selective activation of visual cortex neurons by fixational eye movements: implications for neural coding. Visual Neuroscience, 18(2), 259-277.
Troxler, D. (1804). Uber das verschwinden gegebener gegenstande innerhalb unseres gesichtskreises. Ophthalmologische bibliothek, 2(2), 1- 53.
None declared
Dear Editor,
One of the most controversial issues in current oculomotor research is the influence of microscopic saccades on visibility. To avoid further unnecessary controversy and confusion, I feel compelled to point out numerous methodological and conceptual issues contained in the article by McCamy et al. on this topic.
In brief, McCamy and colleagues show that in highly artificial conditions o...
Dear Editor,
One of the most controversial issues in current oculomotor research is the influence of microscopic saccades on visibility. To avoid further unnecessary controversy and confusion, I feel compelled to point out numerous methodological and conceptual issues contained in the article by McCamy et al. on this topic.
In brief, McCamy and colleagues show that in highly artificial conditions of 30 sec continuous fixation with the immobilized head, subjects' reports of increasing perceived visibility of a low contrast and low spatial frequency Gabor patch tend to be more often preceded by a saccade than reports of decreasing visibility. From this finding, the authors conclude that microsaccades contribute to "restoration" of foveal and peripheral vision during natural viewing.
First, this study does not address "restoration of visibility" or "fading", but perceptual contrast sensitivity. Subjects were asked to continuously report whether the stimulus is "intensifying" or "fading". Such 2-alternative forced choice did not allow reporting what arguably would be the most frequent "no change" percept. Furthermore, it is not clear if the stimulus actually ever faded or was just perceived as less intense, and how the percept differed at different eccentricities. Therefore, the claim that images fade in the fovea, is not substantiated even for the study's specific conditions.
Second, this study does not show any effect for real microsaccades, as they are normally defined in the literature. As seen in the Fig. 7B, only saccades larger than 15' show some perceptual influence, and despite the highly artificial conditions, specifically designed to induce as much "fading" as possible, the actual strength of the effect reveals only a weak relationship between saccades and perceptual changes (only 23% of <1 deg saccades preceded increasing visibility reports).
Third, the large and fuzzy stimulus used in the study is not suited to investigate foveal effects. Fig. 1A is misleading since the size of the small fixation spot (0.05 deg) is illustrated as being nearly equal to the size of the stimulus (~4 deg). The stimulus was also at low spatial frequency and contrast. But stimuli containing fine features, which are what the foveal region evolved to deal with, do not fade, even after prolonged fixation (Collewijn and Kowler, 2008). For example, the fixation spot presumably did not fade; otherwise subjects could not have performed the task.
Fourth, authors say that "microsaccades of all directions are equally effective" in "agreement with previous physiological analyses of microsaccade direction on the firing of V1 neurons", but the earlier paper from this group (Martinez-Conde et al., 2002) shows a large difference between microsaccade-related V1 firing in presence of optimally-oriented vs orthogonal stimulus. Although this study did not quantify the spatial interactions of stimuli and saccade directions, the careful analysis of spatiotemporal interactions between eye movement trajectories, receptive fields and stimuli clearly shows the dependence of neuronal firing patterns on saccade directions and stimulus spatial features (Kagan et al., 2008; Snodderly et al., 2001). The lack of such dependence in the present study is yet another indication that the stimulus was ill-suited to study interactions of eye movements and visual features.
Fifth, the average fixation duration in natural viewing is ~300 ms (i.e. ~3 saccades per sec), and even low contrast peripheral stimuli don't fade during those short bouts of fixation. Fixations of 30 s never occur under natural viewing conditions. In fact, even under highly unnatural conditions of this study, the reports of decreased visibility occurred only every 10-13 s. Even if the authors had found a significant effect for microsaccades in their experiments (but they did not, see above), this finding would have carried little implications for everyday vision.
Sixth, the subjects are effectively required to perform two tasks: report contrast and maintain fixation. It is known that one of the functions of microsaccades is to maintain steady fixation (see Cherici et al., 2012 for most recent data; Engbert et al., 2011 for the model; and Rolfs, 2009 for review). This implies that corrective saccades triggered by retinal slip are, in this study, interpreted as preventing "fading". In normal unconstrained vision, microsaccades and saccades are exploratory, i.e. they bring the fovea to a specific region of interest, and likewise are not triggered by the "fading" of the surrounding world.
Seventh, the analogy equating microsaccades/visibility with the retina/LGN firing is misleading. Unlike retinal activity, which depends on external factors, microsaccades are known to be under oculomotor control (Hafed et al., 2009; Ko et al., 2010). The finding that induced image fading does not lead to an increment in microsaccades (Poletti and Rucci, 2010) lends further support to the idea that microsaccades (or saccades) have nothing to do with image fading.
Due to space limitations, this response does not address other numerous issues contained in the article, such as ignoring the large body of literature against a causal role for microsaccades in image fading, other sources of retinal motions including head and body movements, the underestimated contribution of drift, and the suppression of microsaccades during sustained attention tasks.
Contrary to what authors claim, it is generally accepted that the retinal motion caused by the microsaccades or saccades can restore visibility of a faded stimulus in the laboratory (e.g. Collewijn and Kowler, 2008). The question is whether this is a function of microsaccades during normal everyday viewing. Given that (1) images do not fade in everyday vision at normal fixation durations -- certainly not in the fovea, despite repeated assertions by the authors to the contrary, and (2) that microsaccades tend to be infrequent during normal visual behaviors and (3) even suppressed during sustained attention, preventing image fading cannot be neither a "fundamental" nor an "accidental" function of microsaccades.
The perceptual and neuronal effects of fixational eye movements have indeed been a topic of intense debate and controversy for a long time, in part due to technical limitations of early experiments. A possible role of microsaccades in visibility was a viable hypothesis until the 1980s, when it was discovered that the physiological motion of the head present under natural conditions is already sufficient to ensure that images never fade (work of Steinman and colleagues). Since then, thanks to advances in both experimental and computational techniques, our understanding of the real functional significance of drifts and microsaccades has made significant steps forward (Ahissar and Arieli, 2012; Hafed et al., 2011; Kowler 2011; Kuang et al., 2012; Rucci et al., 2007).
References
Ahissar E, Arieli A (2012) Seeing via miniature eye movements: a dynamic hypothesis for vision. Front. Comput. Neurosci. 6:89.
Collewijn H, Kowler E (2008) The significance of microsaccades for vision and oculomotor control. J Vis 8:20.1-20.21.
Engbert R, Mergenthaler K, Sinn P, Pikovsky A (2011) An integrated model of fixational eye movements and microsaccades. Proc Natl Acad Sci USA 108:E765-E770.
Hafed ZM, Goffart L, Krauzlis RJ (2009) A neural mechanism for microsaccadegeneration in the primate superior colliculus. Science 323:940 -943.
Hafed ZM, Lovejoy LP, Krauzlis RJ (2011) Modulation of microsaccades in monkey during a covert visual attention task. J Neurosci 31:15219- 15230.
Kagan I, Gur M, Snodderly DM (2008) Saccades and drifts differentially modulate neuronal activity in V1: effects of retinal image motion, position, and extraretinal influences. J Vis 8:19.1–19.25.
Ko HK, Poletti M, Rucci M (2010) Microsaccades precisely relocate gaze in a high visual acuity task. Nat Neurosci 13:1549 -1553.
Kowler E (2011) Eye movements: the past 25 years. Vision Res 51:1457- 1483.
Kuang X, Poletti M, Victor JD, Rucci M (2012) Temporal encoding of spatial information during active visual fixation. Cur Biol 22(6): 510- 514.
Martinez-Conde S, Macknik SL, Hubel DH (2002) The function of bursts of spikes during visual fixation in the awake primate lateral geniculate nucleus and primary visual cortex. Proc Natl Acad Sci USA 99:13920-13925.
Poletti M, Rucci M (2010) Eye movements under various conditions of image fading. J Vis 10:6.1- 6.18.
Rucci M, Iovin R, Poletti M, Santini F (2007) Miniature eye movements enhance fine spatial detail. Nature 447:851-854.
Snodderly DM, Kagan I, Gur M (2001) Selective activation of visual cortex neurons by fixational eye movements: Implications for neural coding. Vis Neurosci 18: 259-277.
None declared