| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Report
Department of Neurobiology and Anatomy and Center for Visual Science, University of Rochester, Rochester, New York 14642
Submitted 1 May 2003; accepted in final form 9 June 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
). Visual motion is one of the more important features of objects in the environment, and the mechanisms underlying the processing of visual motion have been extensively studied, both psychophysically and physiologically (for recent review see Pasternak et al. 2003). In contrast, less is known about the way visual motion is stored and subsequently used to guide motor action. A few studies that did examine this question showed that the information about visual motion retained in memory is spatially localized (Zaksas et al. 2001) and can be retained in memory for many seconds with good fidelity (Blake et al. 1997; Magnussen and Greenlee 1992). In this study we focused on the properties of the remembered motion information. We used a novel approach in which we spatially separated the encoding and retrieval/comparison phases of the working memory task. This separation allowed us to use a random-motion mask to examine the attributes of the remembered stimulus as well as to measure the spatial and temporal characteristics of the underlying mechanism. Our results demonstrate that the memory representation is highly specific, retaining the information not only about stimulus direction and its location but also about its size and speed. Furthermore, our results show that this specific representation of visual motion is most fragile immediately after encoding. This faithful preservation of all attributes of the remembered motion stimulus suggests active involvement of motion processing cortical areas in its storage.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
for description). Experiments were carried out in accordance with the guidelines published in the National Institutes of Health Guide for the Care and Use of Laboratory Animals (National Institutes of Health publication No. 86–23, revised 1987). Stimuli and behavioral procedures
STIMULI. The stimuli consisted of moving dots within a stationary circular aperture, 3° or 4° in diameter, repeatedly displaced in a direction of motion chosen randomly from a uniform distribution of directions (Fig. 1A). Such stimuli were first introduced by Williams and Sekuler (1984) and subsequently used extensively by this and other laboratories (e.g., Blake et al. 1997; Pasternak et al. 1990; Watamaniuk et al. 1989). In most experiments, the lifetime of an individual dot was equal to the duration of the stimulus presentation (300 ms). A few measurements (in Monkey 2) were performed with dot lifetime set to 50 ms. The dots were displaced by a constant step size (
x) and temporal interval (t =13 ms) and the dot density was kept at 4.7 dots/deg2. In all experiments, all three stimuli (sample, mask, and test) were always presented at an eccentricity of 7° in the upper or lower portions in the noncorresponding quadrants of the visual field (see Fig. 1B).
|
The range of the distribution of directions set to 0° renders a stimulus with all the dots displaced in the same direction, while the distribution of directions of 360° contains only local random motion of individual dots and no net motion. When the distribution was narrower than about 320–340°, the dots appeared to flow in the direction of the mean of the distribution. The sample stimulus consisted of dots repeatedly displaced in a direction chosen randomly from a uniform distribution of directions (Fig. 1A). The test stimulus always consisted of dots moving coherently (direction range = 0°) in the same or in the opposite direction to that in the sample. Sample and test were either presented at the same spatial location or in opposite, noncorresponding portions of the visual field, but always at equal distance from the fovea (Fig. 1B).
BEHAVIORAL TASK. During each testing session, the monkeys compared the directions of two sequentially presented random-dot stimuli, sample and test, separated by a 1500-ms delay. A tone was presented for the duration of the trial and the monkeys were required to maintain fixation throughout that period. On each trial, the direction of net flow in the sample was chosen at random from eight equally spaced directions and the test moved in a direction that was the same as, or opposite from, that of the sample. The monkeys judged the directions as the same or different by pressing one of two adjacent buttons. Incorrect responses resulted in a 3- to 5-s tone and no reward. The temporal sequence during each trial is shown in Fig. 1C. At various times during the 1500-ms delay, a random-motion stimulus (direction range = 360°), the mask, was introduced either at the location of the sample or at the location of the upcoming test. During each testing session, the location of all stimuli was the same. Thus on each trial the monkeys knew where the sample, the test, and the mask would appear. Each session consisted of 400–800 trials, separated by a 3-s intertrial interval.
THRESHOLD MEASUREMENTS. Performance was evaluated by measuring direction range thresholds in a staircase procedure. In this procedure, three consecutive correct responses resulted in an increase in direction range, while a single incorrect response resulted in a decrease in direction range. The data were fitted with a maximum likelihood Weibull function (Weibull 1951) and the threshold was defined as the stimulus value at which the animal performed at 75% correct. Three to five or more threshold determinations were performed for each stimulus condition. The significance of the effects of each manipulation was determined using a one-way ANOVA.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Spatial localization of the remembered sample
We found that the mask interfered with performance under a limited set of spatial and temporal conditions. The data in Fig. 2 show that the mask had a marked detrimental effect on performance only when it was placed at the location of the upcoming test. Maximal disruption was achieved when the mask was present at that location for 
250–300 ms. At shorter durations, the effect of the mask in the test location was much less pronounced. On the other hand, the effect of the mask at the sample location was minimal, irrespective of its duration.
|
We also examined the extent to which this representation is confined to the test location by placing the mask at several distances from the upcoming test. The data in Fig. 2B show that the mask lost its detrimental effect if it were displaced only a few degrees from the location of the upcoming test. This spatial specificity of the mask effect provides new evidence supporting the notion of spatial localization of the remembered stimulus proposed by Zaksas et al. (2001). That study demonstrated spatial specificity by manipulating the spatial separation between the sample and the test. In the present study, by manipulating the location of the mask we were able to more precisely pinpoint the site of the remembered stimulus to the location where the perceptual decision was to take place.
What is remembered?
To learn more about the nature of the sample representation in memory we determined whether, in addition to motion direction, the information about its size and speed was also retained. We examined the properties of the remembered stimulus by manipulating the size and the local speeds of the mask and using it as a probe.
We found that the mask was most disruptive when its size approximated the size of the remembered sample (Fig. 3), suggesting that the size of the sample was likely to have been preserved in memory.
|
To determine whether the speed of the sample was also preserved in memory, we manipulated the spatial displacement of individual dots (x) in the mask. This manipulation resulted in a mismatch between the local speeds in the mask and the local speeds of the sample. A detrimental effect of the mask despite the mismatch in temporal properties would be indicative of the sample speed not being retained. The results in Fig. 4 show that when sample and test moved at 10°/s (x = 0.13, t = 13 ms), the effect of the mask was most pronounced if the dot displacement in the mask and the sample matched precisely.
|
When is the mask most effective?
To determine the point in time when the representation of the remembered sample is most susceptible to disruption, the random-motion mask was introduced at various points of the delay. The results of these measurements are shown in Fig. 5. In both monkeys, the masking was most pronounced when the mask was introduced into the test location about 100–150 ms after the start of the delay.
|
The effectiveness of the mask decreased later in the delay and by 200–300 ms into the delay thresholds with the mask were nearly identical in sample and in test locations, and similar to those measured without the mask. This extreme temporal specificity of the mask effect suggests that very early in the delay, about 100 ms after the offset of the sample, the neural representation of the stimulus has already been moved to the test location and is most susceptible to disruption.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Interfering effect of the mask
The specific nature of the mask effect is indicative of interference with the remembered stimulus rather than the nonspecific disruption in behavior due to distraction. If this effect was due to simple confusion we would expect to see a decrease in performance that was not related to the nature of the mask or to the remembered sample. Furthermore, the effect of the mask was absent when the remembered stimulus moved coherently (0° range)(Zaksas et al. 2001).
Temporal dynamics of storage
In this experiment the location and the time of appearance of the sample, the mask, and the test were highly predictable. Thus the monkeys knew where and when the mask and the test were going to appear. The spatially selective effect of the mask suggests that, in anticipation of the test appearing in a remote location, the monkeys transferred information about the remembered stimulus to that location. Particular susceptibility of performance to the mask during a relatively narrow temporal window early in the delay points to the possible time of this transfer and to the vulnerability of memory representation early in the delay. We speculate that, around the time of transfer, the representation of the sample in memory may be less stable and thus more susceptible to noise introduced by the mask. It is interesting that effectiveness of the mask appears to be most pronounced at the time of its introduction and is less dependent on its duration (see Fig. 2). It remains to be seen whether the vulnerability of performance 100–150 ms after the start of the delay is the result of neural events taking place at time of mask onset or is related to the time it takes for the mask to reach its maximal effect or both.
The idea of signal instability early in the delay is supported by a recent network model of working memory (Compte et al. 2000; Wang 2001). In this model, sensory information is fed into a network of neurons designated for parametric storage by way of stable recurrent activity. Wang and his colleagues have shown that a distracter signal had its maximal effect when it varied little from the original stimulus along the parameter being retained by such recurrent networks (Compte et al. 2000). Furthermore, it takes a relatively long time for these recurrent circuits to achieve stability because they depend on relatively slow receptor dynamics and on inhibitory circuits with time constants on the order of several hundred milliseconds (Compte et al. 2000; Wang 2001). The properties of the proposed recurrent networks are consistent with our data, which show the maximal disruption by a noisy mask occurring very shortly after the sensory signal in the task-relevant spatial location. A working memory signal therefore appears to be disrupted by parameter-specific noise injected into the system prior to stabilization.
Our data show that the mask had to endure for 200 ms to maximally disrupt performance. This may be indicative of the relatively long temporal integration characteristic of the mechanisms responsible for processing complex motion (Watamaniuk and Sekuler 1992). This observation further strengthens the notion that mechanisms involved in processing of motion stimuli also participate in retaining these stimuli in memory (Bisley and Pasternak 2000; Fuster 1997).
Nature of the remembered stimulus
We used the random-motion mask to reveal specific characteristics of the remembered stimulus. Masking stimuli have previously been used in a number of experiments examining perceptual memory. In these "memory masking" experiments the mask was applied while subjects performed delayed discriminations of speeds (Magnussen and Greenlee 1992) and of spatial frequencies (Bennett and Cortese 1996; Lalonde and Chaudhuri 2002; Magnussen et al. 1991). The masks in these experiments consisted of stimuli that bore different degrees of similarity to the stimuli being discriminated. The general finding of these experiments was that the less the mask resembled the discriminated visual stimuli, the greater the detrimental effect of the mask. Since these masking effects were selective for specific stimulus dimensions, it was concluded that perceptual memory for different stimulus attributes is served by parallel, higher-order perceptual mechanisms (e.g., motion, spatial frequency) (Magnussen and Greenlee 1999).
In our experiments, the mask contained only random motion and thus provided no motion information relevant to the task. It was used to inject noise into the process underlying the retention of the sample (Ball and Sekuler 1979). We reasoned that a detrimental effect of such a "memory mask" would be a reflection of its interference with the representation of the remembered stimulus. Indeed in the previous study we used such masks to demonstrate that this representation is spatially localized to the site of the upcoming test (Zaksas et al. 2001). In that paper we also provided evidence that the sample containing a broad range of directions was more likely to be retained as a veridical representation of the stimulus than a simple vector representing the mean direction. The present findings provide strong support for this notion. The dependence of the mask's effects on its size and local speed suggests that the representation of the sample in memory retains not only the information about its direction as required by the task, but also about its size and speed.
One of the striking observations of our study is the speed-dependent specificity of the mask effect. When the sample moved at a lower speed, the mask was effective only when its local speeds (i.e., x) matched precisely the local speeds of the remembered sample. Together with other reports of visual masks selectively interfering with sensory information (Ball and Sekuler 1979; Bennett and Cortese 1996; Magnussen and Greenlee 1992), these results suggest that the storage of visual signals is likely to be accomplished, at least in part, by the very mechanisms that are involved in their encoding.
Possible neural mechanisms
A number of recent studies from this laboratory have implicated cortical middle temporal (MT) area in the performance of the memory for motion task. Bisley and Pasternak (2000) have shown that deficits produced by lesions of MT and middle superior temporal area (MST) MT//MST were greatest at longer delays separating sample and test stimuli. In a subsequent study, Bisley et al. (2001) reported a severe disruption in performance of the task by microstimulation of MT during the delay. Furthermore, in a recent psychophysical study, we have demonstrated that the spatial scale of the mechanisms underlying the performance of our task closely matches the spatial scale of MT neurons (Zaksas et al. 2001).
The involvement of MT neurons in remembering visual motion is also supported by our recent recordings from MT neurons carried out in the same animals performing the task identical to that used in the present study (Bisley, Zaksas, Droll, and Pasternak, unpublished data). These recordings revealed a pattern of activity early in the delay that may be related to the phenomenon observed here. Specifically, many MT neurons showed transient activation during the first few hundred milliseconds of the delay and the properties of this activation reflected the direction of the preceding sample. While it remains to be seen whether this activation is linked directly to the ability to remember visual motion, its existence in the absence of a visual stimulus implicates area MT in the circuitry subserving memory for visual motion. The striking parallel between the time of maximal effectiveness of the mask and the appearance of early activation in MT suggests a possibility that the mask may be interfering with this early delay activity in MT. While we do not have direct evidence to support this possibility, the present study strongly implicates cortical area(s) with machinery to process complex visual motion in the ability to remember it.
|
DISCLOSURES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprints and other correspondence: T. Pasternak, Department of Neurobiology and Anatomy, Box 603, University of Rochester, Rochester, NY 14642 (E-mail: tania{at}cvs.rochester.edu).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Ball K and Sekuler R. Masking of motion by broadband and filtered directional noise. Percept Psychophys 26: 206–214, 1979.[Web of Science]
Bennett PJ and Cortese F. Masking of spatial frequency in visual memory depends on distal, not retinal, frequency. Vis Res 36: 233–238, 1996.[Web of Science][Medline]
Bisley JW and Pasternak T. The multiple roles of visual cortical areas MT/MST in remembering the direction of visual motion. Cereb Cortex 10: 1053–1065, 2000.
Bisley JW, Zaksas D, and Pasternak T. Microstimulation of cortical area MT affects performance on a visual working memory task. J Neurophysiol 85: 187–196, 2001.
Blake R, Cepeda NJ, and Hiris E. Memory for visual motion. J Exp Psychol Hum Percept Perform 23: 353–369, 1997.[Web of Science][Medline]
Compte A, Brunel N, Goldman-Rakic PS, and Wang XJ. Synaptic mechanisms and network dynamics underlying spatial working memory in a cortical network model. Cereb Cortex 10: 910–923, 2000.
Fuster JM. Network memory. Trends Neurosci 20: 451–459, 1997.[Web of Science][Medline]
Lalonde J and Chaudhuri A. Task-dependent transfer of perceptual to memory representations during delayed spatial frequency discrimination. Vis Res 42: 1759–1769, 2002.[Web of Science][Medline]
Magnussen S and Greenlee MW. Retention and disruption of motion information in visual short-term memory. J Exp Psychol Learn Mem Cogn 18: 151–156, 1992.[Web of Science][Medline]
Magnussen S and Greenlee MW. The psychophysics of perceptual memory. Psychol Res 62: 81–92, 1999.[Web of Science][Medline]
Magnussen S, Greenlee MW, Asplund R, and Dyrnes S. Stimulus specific mechanisms of visual short-term memory. Vis Res 31: 1213–1219, 1991.[Web of Science][Medline]
Pasternak T, Albano JE, and Harvitt DM. The role of directionally selective neurons in the perception of global motion. J Neurosci 10: 3079–3086, 1990.[Abstract]
Pasternak T, Bisley JW, and Calkins D. Visual information processing in the primate brain. In: Biological Psychology, edited by Gallagher M and Nelson RJ. New York: Wiley, 2003, chapt. 6, p. 139–185.
Rudolph K and Pasternak T. Transient and permanent deficits in motion perception after lesions of cortical areas MT and MST in the macaque monkey. Cereb Cortex 9: 90–100, 1999.
Wang XJ. Synaptic reverberation underlying mnemonic persistent activity. Trends Neurosci 24: 455–463, 2001.[Web of Science][Medline]
Watamaniuk SN and Sekuler R. Temporal and spatial integration in dynamic random-dot stimuli. Vis Res 32: 2341–2347, 1992.[Web of Science][Medline]
Watamaniuk SN, Sekuler R, and Williams DW. Direction perception in complex dynamic displays: the integration of direction information. Vis Res 29: 47–59, 1989.[Web of Science][Medline]
Weibull W. A statistical distribution function of wide applicability. J Appl Mech. 18: 292–297, 1951.
Williams DW and Sekuler R. Coherent global motion percepts from stochastic local motions. Vis Res 24: 55–62, 1984.[Web of Science][Medline]
Zaksas D, Bisley JW, and Pasternak T. Motion information is spatially localized in a visual working-memory task. J Neurophysiol 86: 912–921, 2001.
This article has been cited by other articles:
|
|
 |
|
  E. P. Merriam and C. L. Colby Active Vision in Parietal and Extrastriate Cortex Neuroscientist, October 1, 2005; 11(5): 484 - 493. [Abstract] [PDF]
|
|
|
|
 |
|
 S. J. Bennett and G. R. Barnes Predictive Smooth Ocular Pursuit During the Transient Disappearance of a Visual Target J Neurophysiol, July 1, 2004; 92(1): 578 - 590. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
