E-Letter responses to:
Ziad M. Hafed, Laurent Goffart, and Richard J. Krauzlis
A Neural Mechanism for Microsaccade Generation in the Primate Superior Colliculus
Science 2009; 323: 940-943
Martin Rolfs
Laboratoire Psychologie de la Perception
Université Paris Descartes
45, rue des Saints-Pères
75006 Paris, France
Z. M. Hafed et al. show in their Report ("A neural mechanism for microsaccade generation in the
primate superior colliculus," 13 February 2009, p. 940) how the brain generates the smallest rapid
eye movements. Rostral neurons of the superior colliculus (SC) encode goal locations very close
to the fovea and fire vigorously before microsaccades into their response field. The authors have
made an important step in revealing the causal evidence for this mechanism, which had previously
been explained in theory based principally on psychophysical evidence (1). Understanding the
origin of fixation instability illuminates the basic limits of visual perception; in the absence of retinal
image motion, perception fades (2, 3). However, fixation instability comprises both microsaccades
and the slower, erratic drift movements (4), and both may work together to provide sufficient retinal
image slip (5). I therefore highlight two important implications of this research for understanding
how fixation instability is achieved to serve visual function.
First, if fixation instability serves to provide a necessary baseline of retinal motion, then a dynamic
visual input may reduce the level required for the baseline. Indeed, one of the hallmarks of fixation
behavior is a strong drop in microsaccade frequency in response to sensory events, an inhibition
that may result from a strong decrease of rostral SC activity (1), which is compatible with Hafed et
al.'s findings.
Second, the SC is involved in generating slow eye movements that pursue a moving target (6, 7).
Therefore, I propose that if an imbalance of activity in the rostral SC does not generate a
microsaccade, it biases slow drift movements, accounting for the cooperation of both in fixation
control (8, 9). Examining the relation between spontaneous drift movements and rostral SC activity
is certainly more of a challenge than examining the relation for microsaccades. However, this
would provide the cornerstone for our comprehension of overall fixation control.
Martin Rolfs
References
1. M. Rolfs, R. Engbert, R. Kliegl, J. Vis. 5, 11:5 (2008).
2. R. W. Ditchburn, B. L. Ginsborg, Nature 170, 36 (1952).
3. L. A. Riggs, F. Ratliff, J. Opt. Soc. Am. 42, 872 (1952).
4. F. Ratliff, L. A. Riggs, J. Exp. Psychol. 100 687 (1950).
5. R. Engbert, K. Mergenthaler, Proc. Natl. Acad. Sci. USA 103, 7192 (2006).
6. R. J. Krauzlis, M. A. Basso, R. H. Wurtz, Science 276, 1693 (1997).
7. M. A. Basso, R. J. Krauzlis, R. H. Wurtz, J. Neurophysiol. 84, 892 (2000).
8. J. Nachmias, J. Opt. Soc. Am. 49, 901 (1959).
9. R. Engbert, R. Kliegl, Psychol. Sci. 15, 431 (2004).
A Neural Mechanism for Fixation Instability

Ziad M. Hafed, Richard J. Krauzlis
Systems Neurobiology Laboratory
Salk Institute for Biological Studies
10010 North Torrey Pines Road
La Jolla, CA 92037, USA
Laurent Goffart
Institut de Neurosciences Cognitives de la Méditerranée
Equipe Dynamique de la Perception Visuelle et de l Action, UMR 6193
CNRS–Aix-Marseille Universités
13402 Marseille, France.
We agree that the functional role of microsaccades and other fixational eye movements remains
an interesting open question. We take issue, however, with M. Rolfs' use of the term "fixation
instability." In our Report, we did not uncover a "neural mechanism for fixation instability," but rather
a mechanism for generating a type of eye movement that is frequently observed during gaze
fixation. There are several pathological conditions that involve problems with maintaining stable
fixation, such as square-wave jerks, saccade oscillations, and ocular flutter (1). We suggest that
the term "fixation instability" be reserved for these clinically identified oculomotor disturbances, and
not for what is apparently a normal aspect of visuomotor function.
The questions of what triggers microsaccades, why their frequencies may be modulated by
stimulus transients, and how visual functions are dependent on or utilize them are still
unanswered. For example, with respect to the idea that microsaccades are important for vision
because they prevent retinal image fading, many other contributors to retinal image motion do
exist, including movements of the head, body, and objects in the world. In addition, high-acuity
visual tasks that require focused and foveal attention, such as threading a needle (2), result in a
reduction in microsaccade frequency and therefore a reduction in retinal image motion, not the
opposite. Thus, it may be the case that the importance of fixational eye movements for visual
perception lies more in how they refine the spatial localization of minute visual features or how they
modulate the image statistics of natural scenes (3) than in how they refresh retinal images. These
questions should continue to be a topic of active scientific inquiry.
We also agree that the neural mechanisms for slow drifts, and the functional role of these
movements, deserve further investigation. Rolfs alluded to previous results relating the superior
colliculus (SC) to smooth pursuit eye movements (4–7), and he suggested similar involvement in
slow drifts. We should clarify here that these previous results have shown that SC activity is
important for identifying the location of the goal (8) for pursuit, which is distinct from providing a
premotor signal for the generation of slow eye movements. It is therefore possible that modulations
of SC activity during slow drifts reflect the continuously shifting locus of the fixated target caused
by these small eye movements. It is also possible that such modulations, if present, reflect
ascending inputs (9) to the SC about eye position. In either case, the neural control of slow drifts is
expected to arise from multiple distributed brain circuits, which may or may not include the SC.
There are also several other important questions that our study raises and are worthy of being
mentioned here and investigated in the future, both at the behavioral and neurophysiological
levels. Some of these questions are: Does the neural control of microsaccades originate in the SC,
or are there inputs from the frontal eye fields (FEF) or other cortical areas that play a crucial role in
how and when these movements are generated? Are there corollary discharge signals for
microsaccades sent to cortex, as there are for larger voluntary saccades? Are microsaccades and
Reply to M. Rolfs' E-Letter

other fixational eye movements actively controlled by the brain to condition how visual signals are
encoded? Answering these, and other questions about foveating mechanisms promises to be a
very exciting challenge.
References
1. R. J. Leigh, D. S. Zee, The Neurology of Eye Movements (Oxford University Press, Oxford, ed.
4, 2006).
2. B. J. Winterson, H. Collewijn, Vision Res. 16, 1387 (1976).
3. M. Rucci, R. Iovin, M. Poletti, F. Santini, Nature 447, 851 (2007).
4. R. J. Krauzlis, M. A. Basso, R. H. Wurtz, Science 276, 1693 (1997).
5. M. A. Basso, R. J. Krauzlis, R. H. Wurtz, J. Neurophysiol. 84, 892 (2000).
6. Z. M. Hafed, L. Goffart, R. J. Krauzlis, J. Neurosci. 28, 8124 (2008).
7. Z. M. Hafed, R. J. Krauzlis, J. Neurosci. 28, 9426 (2008).
8. R. J. Krauzlis, D. Liston, C. D. Carello, Vision Res. 44, 1445 (2004).
9. R. Hartwich-Young, J. S. Nelson, D. L. Sparks, Vis. Neurosci. 4, 29 (1990).