HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: 94/6/4094    most recent 00303.2005v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (10) Citing Articles via Google Scholar Google Scholar Articles by Chapman, C. E. Articles by Meftah, E.-M. Search for Related Content PubMed PubMed Citation Articles by Chapman, C. E. Articles by Meftah, E.-M.

Independent Controls of Attentional Influences in Primary and Secondary Somatosensory Cortex

¹Département de Physiologie, Groupe de Recherche sur le Système Nerveux Central, and ²École de Réadaptation, Faculté de Médecine, Université de Montréal, Montreal, Quebec, Canada

Submitted 22 March 2005; accepted in final form 24 August 2005

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
The neuronal mechanisms underlying enhanced perception of tactile stimuli with directed attention were investigated using single-unit recordings from primary (S1, n = 53) and secondary (S2, n = 50) somatosensory cortex in macaque monkeys. Neuronal responses to textures scanned under the digit tips (spatial periods, SP, of 2, 3.7 or 4.7 mm) were recorded while attention was directed either to discriminating a change in texture or to the reward and also in a neutral no-task condition. Cell discharge was quantified in three periods of the trials: salient texture (directed attention), postreward, and static (both cases, attention directed to the reward). S1 texture- and non-texture-sensitive cells, as well as S2 non-texture-sensitive cells, showed a modest enhancement of discharge during the salient texture period (25%) but no change in response gain, consistent with an additive increase in neuronal responsiveness with directed attention. In contrast, S2 texture-related cells showed a larger enhancement with directed attention to salient inputs (82%) and increased response gain, suggesting that directed attention produces a multiplicative increase in S2 responsiveness. During the postreward period, and also in no-task testing, S1 texture-sensitive cells preserved their sensitivity to SP. In contrast, S2 texture-, but not non-texture-, sensitive cells showed a marked suppression of discharge and decreased gain after the discrimination response. Together, the results support the notion that S2 discharge reflects stimulus parameters in relation to ongoing behavioral demands. The results also support the existence of two independent attentional mechanisms in somatosensory cortex, one generalized (S1 and S2), and the other focused on S2 texture-related cells.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Many factors can intervene to modulate cortical neuronal responsiveness to sensory stimuli, including attention and movement (Chapman 1994; Desimone and Duncan 1995; Kanwisher and Wojciulik 2000; Kastner and Ungerleider 2000). Parallel changes in perception have also been reported, suggesting a causal link between neuronal responsiveness and perception (Ress et al. 2000; Spitzer and Richmond 1991; Spitzer et al. 1988; Williams et al. 1998).

Much of our knowledge about the effects of directed attention on the processing of sensory stimuli comes from studies in the visual system. Taken together, evidence indicates that attention can modulate neuronal responses to visual stimuli at early stages in the visual processing pathways (e.g., primary visual cortex, V1), but that the effects are more prominent in later stages of processing (V4, inferotemporal cortex) (reviewed in Desimone and Duncan 1995; Kanwisher and Wojciulik 2000). One of the principal findings is that discharge rates are strongly modified by directed attention. The underlying mechanisms include gating, or filtering, of non-attended inputs (Moran and Desimone 1985), changes in baseline firing rates (Luck et al. 1997), and increases in neuronal responsiveness to attended stimuli with (Reynolds et al. 2000; Spitzer et al. 1988) or without (McAdams and Maunsell 1999; Treue and Martinez Trujillo 1999) increases in response selectivity (sharpening of tuning curves, leftward shift in the stimulus-response function).

In contrast, much less is known about the effects of directed attention on the processing of somatosensory stimuli, specifically tactile stimuli. It is only relatively recently that selective attention has been shown to unequivocally enhance the perception of tactile stimuli (Driver and Spence 1998; Post and Chapman 1991; Zompa and Chapman 1995). Consequently, there have been few studies of the underlying neuronal mechanisms. However, we do now know that directed attention can modify neuronal responsiveness to tactile stimuli in both primary (S1) and secondary (S2) somatosensory cortex (Burton and Sinclair 2000; Burton et al. 1997; Hsiao et al. 1993; Meftah et al. 2002; Steinmetz et al. 2000). In addition, attentional effects are more frequent, more robust, and more complex in a higher order area, S2, as compared with S1 (Hsiao et al. 1993; Meftah et al. 2002).

In our recent study using a cross-modal manipulation of attention, tactile versus visual (Meftah et al. 2002), we proposed two alternate hypotheses to explain these observations—independent controls of parietal cortical discharge exerted at each level, S1 and S2, or top-down controls from S2 to S1 (see also Burton et al. 1999); however, examination of this issue was outside the scope of our previous publication. If attentional controls over somatosensory responsiveness in S1 and S2 cortex are exerted independently at each level in processing, then one would expect to see systematic differences in the manner in which attention or stimulus salience modulates, for example, response selectivity in the two areas. If, on the other hand, the attentional effects seen at early stages of processing (S1)—being weaker and less frequent—simply reflect top-down controls from higher level stages (S2), and this in a manner consistent with the known anatomical connectivity (Friedman et al. 1986), then one would expect that the underlying neuronal mechanisms would be similar at both levels.

The purpose of this study was to explore these two possibilities, using the same behavioral task as in our previous study. We addressed three questions in this report. 1) Are the stimulus-response functions of texture-sensitive neurons modified in a similar manner by attention in both S1 and S2? 2) Does attention enhance responses to attended stimuli or, conversely, suppress responses to non attended inputs? 3) Are all cells that are activated by the tactile stimulation, including those whose discharge covaried with SP (texture-sensitive) and those that did not (non-texture-sensitive), modulated in a similar fashion by attention in both areas?

The influence of attention on SP sensitivity was addressed by quantifying neuronal discharge frequency in relation to three SPs (2, 3.7, and 4.7 mm) scanned under the immobile digit tips of monkeys under three different contexts: attention directed to the salient change in texture (salient texture period of a texture discrimination task), attention directed to the reward (postreward period of task trials), and a neutral condition (no-task). By using the discharge in the neutral condition as the reference value, we determined the sign of the modulation associated with directed attention. Finally, we investigated the distribution of the attentional influences in S1 and S2 by comparing the pooled results of two groups of cells in each area: those that were defined as texture sensitive during the salient texture period, and those that were non-texture sensitive during the same period. Taken together, this study addresses the basic neuronal mechanisms that underlie the enhancement of perception of tactile stimuli with directed attention. A preliminary report of the results has been presented (Meftah and Chapman 2003).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Behavioral tasks

Experiments were performed in the same two adult monkeys (Macaca mulatta; G, 8.5 kg; I, 9.2 kg) as used in our previous study (Meftah et al. 2002). Details of the behavioral tasks, data acquisition, histological controls, and the general characteristics of the database were previously described. In brief, monkeys were trained to perform two discrimination tasks, tactile and visual. Each trial was preceded by an instruction cue (green or red light, respectively) that signaled the modality to be discriminated on the upcoming trial. During data acquisition, the tactile and visual tasks were interleaved, so that the monkey had to direct its attention to the appropriate modality on a trial-by-trial basis.

During the recording session, the animal was seated in a primate chair, with the tactile stimulator (a cylindrical drum of 40-cm circumference) firmly clamped to the chair at waist height (see Fig. 1 in Meftah et al. 2002). A raised-dot surface (single continuous band, 2 x 40 cm, containing the standard and modified textures) was affixed around the circumference of the drum. Access to the surface was through a small rectangular aperture. The animals were conditioned to position their digit tips (D3/4) on the surface that formed the floor of the aperture. A response lever was mounted in front of the nonstimulated hand. A light display was mounted directly in front of the monkey at eye level (35 cm distance). This contained the yellow light that served as the visual stimulus [3 x 3 array of yellow light-emitting diodes (LEDs)], and the two instruction lights (green and red LEDs).

The sequence of events in the tactile discrimination task is shown in Fig. 1 (top). During the inter-trial interval (not shown), the drum was repositioned to the start of the segment to be presented in the upcoming trial (2 mm SP in all cases). A brief tone signaled that the surface was in position. Each trial was then self-initiated: 500 ms after the monkey depressed a lever with the nonstimulated hand, a green instruction light was illuminated. Two seconds later, the standard tactile and visual stimuli were presented. The standard texture had a SP of 2 mm (rectangular array of raised dots, dot spacing constant within and between rows); this was displaced under the digit tips (direction of scan, proximal to distal) at 50 mm/s. The standard visual stimulus (10.6 cd/m²) was clearly visible under the recording conditions. The digits encountered the modified surface [dot spacing between rows incremented to 3.7 or 4.7 mm, corresponding to a subjective increase in roughness (Meftah et al. 2000)] 1 or 1.7 s after the onset of the scan. Monkeys were rewarded with a drop of juice for detecting the change in surface texture by releasing the response lever. At the end of the stimulation period (3 s total duration), drum rotation stopped; the visual stimulus and the instruction cue were also extinguished. Data acquisition continued for an additional 0.7 s to provide a baseline value as the monkey sat quietly with its digits in contact with the now immobile surface.

View larger version (34K): [in this window] [in a new window]   FIG. 1. Schematic representation of the sequence of events in the tactile and visual discrimination tasks. Mean firing rate was measured in 3 intervals (shaded columns). Period 1, salient (change) texture, comprised 2 intervals, a = standard 2 mm spatial period (SP); b = modified surfaces (3.7 or 4.7 mm SP). Period 2, postreward, corresponded to the final 400 ms of surface scanning (SPs of 2, 3.7, and 4.7 mm). Period 3, static, included the final 400 ms of the trial (all 3 SPs). For the tactile task, measures for the 2 mm SP were only obtained during period 1a; for periods 2 and 3, measures for the 2 mm SP came from the visual trials (see METHODS). Data for period 1, a and b, of the visual task were measured as shown for the tactile task, but were restricted to SPs of 2 and 4.7 mm (prior to the light change and lever release).

For the visual discrimination task, the sequence of events was similar (Fig. 1, bottom). In this case, the instruction cue was a red light. The standard visual stimulus (above) was modified by increasing luminance to 27.1, 43.9, or 60.8 cd/m², at one of three delays after the onset of the standard stimulus, 0.9, 1.3, or 1.7 s. As for the tactile task, both modalities changed in 50% of the trials; the change in texture either preceded (paired with long visual delay) or followed (paired with short visual delay) the change in light intensity. Again, response time had to fall within a precise time window after the change in light intensity, ensuring that only responses to the visual change were rewarded.

Animal care and housing conformed with published guidelines (Canadian Council on Animal Care, National Institutes of Health), and the experimental protocol was approved by the institutional ethics committee.

No-task condition

Responsiveness to the textured surfaces was assessed outside of the context of the tactile and visual discrimination tasks when possible (cell isolation maintained). For these recordings, the light display was removed (no instruction cue, no visual stimulus) as was the response lever. Using the same stimulation time course as in the tactile and visual tasks, the surfaces were displaced under D3/4 at 50 mm/s while random drops of juice were presented to the animal. The monkey was given two or three drops of juice during each 5-s trial interval, as well as in the intertrial interval, and there was no particular relation between the timing of the juice drop and the time of the texture change (could occur before and/or after). The monkeys appeared to be alert throughout these recordings, eyes open and directed to the juice spigot immediately in front of the animal.

Recordings

Once training was complete (low and stable error rate, <10%), extracellular activity of single neurons in S1 and S2 cortex was recorded with a glass-coated tungsten microelectrode. The locations of the penetrations for one monkey (G) have been published (Fig. 4 in Meftah et al. 2002); histology is not yet available for the other monkey (I), but the recordings confirmed the results obtained in monkey G in all respects. Complete testing of each cell (53 in S1, 50 in S2) in the task condition required 120 trials: 60 tactile and 60 visual trials (order quasi-random). In 72 cells, an additional series of recordings were carried out in the no-task condition (minimum 10 trials for each SP).

View larger version (31K): [in this window] [in a new window]   FIG. 4. Representative S2 texture-sensitive neuron (same cell as in Fig. 2B) showing texture-sensitivity during the salient texture period (1) and dramatically decreased firing rates during the postreward period (2) with loss of texture-sensitivity. In the no-task condition (A, open histograms; B, thin lines), texture sensitivity was observed in both dynamic periods (1 and 2), but firing rates differed markedly from those seen in the task recordings, consistent with a relative enhancement of salient inputs and active suppression of responsiveness after the reward (and extending into the static period) when the input was no longer relevant. Data plotted as in Fig. 3.

The task and the data acquisition were under computer control. The following data were collected for each trial: neural spike intervals (1-ms resolution), vertical contact force (digitization rate, 200 Hz), cue condition, and the time of various events in the trial. For each trial, we carefully monitored the position of the stimulated digits, rejecting trials in which the monkey did not maintain contact with the surface throughout the trial. We also inspected the vertical contact force records, rejecting trials in which contact force varied by more than ±0.2 N prior to the time of the lever response.

For the data from the tactile and visual discrimination tasks (hereafter referred to as the task condition), cell discharge frequency and contact force were measured in three periods in each trial as the animal performed the tactile and visual tasks. The duration of the analysis intervals was a compromise between maximizing the duration to ensure multiple repetitions of the spatial patterns, and respecting task constraints (i.e., minimal reaction time for the lever release). As shown in Fig. 1, period 1 contained the salient texture (trials from the tactile discrimination task); this interval preceded the animal’s discrimination response (lever release). Period 1 was subdivided into two 200-ms intervals, a and b. Period 1a corresponded to the final 200-ms presentation of the standard surface (2 mm SP). Period 1b corresponded to the initial 200 ms of discharge evoked by the modified surfaces (3.7 or 4.7 mm SP). Burst onset was measured in each trial, using an interactive program, and corresponded to the first discharge that followed digit contact with the modified surface as determined by inspecting the individual force traces. At very high resolution, individual force traces showed a periodic fluctuation as each row of raised dots contacted the fingertips, and the period increased when the fingertips contacted the modified surface (increased SP). [Note that this analysis, adjusting the window on a trial-by-trial basis, represented a change from our previous approach whereby firing rate was calculated in relation to the time that the texture change first entered the aperture (Meftah et al. 2002). This was justified as our original method included a period of contact with the standard surface and so tended to minimize the changes in firing rate elicited by the modified surface in each task condition]. Period 2, postreward, followed the animal’s discrimination response, and corresponded to the final 400 ms of the texture presentation (all SPs and all trials). Period 3, static, encompassed the final 400 ms of the trial (300–700 ms after the end of the texture presentation; all SPs and all trials). The texture was moving under the finger tips for periods 1 and 2 (dynamic stimulation); the texture was immobile in the static period. Only data from rewarded trials were included in the analysis, thus ensuring that attention was directed to the tactile stimulus (period 1) or to the reward (periods 2 and 3). For the latter periods, data from the tactile and visual discrimination tasks were pooled after first ensuring that discharge frequency, for the same SP, did not vary as a function of the task (independent t-test, P 0.01; compare the black and red lines in Fig. 2). This was justified for the 92% of cells that showed no difference; the remaining 8% were excluded from these analyses. Note, however, that data for the standard surface (2 mm SP) in periods 2 and 3 came only from the visual discrimination task trials. For the no-task condition, discharge frequency was measured in the same periods. In this case, however, the first period, early, was not salient; and the second period, late, was not systematically preceded by a reward.

View larger version (40K): [in this window] [in a new window]   FIG. 2. Mean firing rate and vertical contact force during the tactile (black line) and visual (red line) discrimination tasks of representative S1 (A) and S2 (B) cells, along with their receptive field. The S2 cell, but not the S1 cell, was previously classified as attention sensitive (tactile > visual during the instruction, standard and salient texture periods). The modified surface had a 4.7 mm SP in these trials. Note that the texture was not salient during the visual discrimination task. c, contralateral; D3/4, digits 3 and 4; LT, light touch; RA, rapidly adapting; SA, slowly adapting.

For cells tested in both the task and no-task conditions, the degree of modulation in discharge associated with each SP during periods 1 and 2 was quantified by calculating a modulation index, MI = (task – no-task)/(task + no-task). In contrast to the linear regression analysis, this approach required no assumption as to the linearity of the relation between cell discharge frequency and SP. The results (period 1 only) were compared with the attention modulation index used previously, AMI = (tactile task – visual task)/(tactile task + visual task) (Meftah et al. 2002).

Repeated-measures ANOVAs were employed for population analyses, specifically to compare discharge frequencies across periods, SPs and task conditions. Post hoc contrast analyses (univariate F-tests) were employed for specific comparisons. The ² test of independence was used for comparisons of frequencies. To determine the contribution of contact force to the changes in firing rate, this was included as a covariate in ANOVAs that included period, SP and, in some cases, task condition, as main factors. To ensure that differences in firing rates between cells did not contribute overly to the results, the latter analysis was repeated after normalizing cell discharge rates, dividing the discharge rate for each trial by the averaged value of all trials (3 periods, 3 SPs and 2 task conditions). All analysis was done with Systat Version 9.0 for Windows (SPSS, Chicago, IL).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
Neuronal data were collected in two monkeys (M. mulatta) trained to perform a task involving a cross-modal manipulation of attention (tactile vs. visual). These data are a subset of the S1 (areas 3b, 1 and 2) and S2 cells used in the study of Meftah et al. (2002). The initial data set consisted of cells recorded in the cutaneous representation of the hand of the hemisphere contralateral to the stimulated hand, S1 (n = 102; areas 3b, 1 and 2) and S2 (n = 76). All cells had a cutaneous receptive field that included the distal phalanges of D3/4, and their discharge was significantly modulated during the presentation of the combined tactile and visual stimuli as compared with the discharge at rest. This report concentrates on 35/102 S1 cells and 36/76 S2 cells that were classified as texture-sensitive during period 1 of the texture-discrimination task (examples shown in Fig. 2), i.e., their discharge frequency was significantly different for the modified surfaces (3.7 and 4.7 mm SP), as compared with the standard surface (2 mm SP), during the salient texture period. For comparison, we present data from non-texture-sensitive cells (18 in S1, 14 in S2) that met the following criteria: a significant increase in discharge during the combined tactile and visual stimulation, no difference in discharge during periods 2 and 3 (tactile vs. visual), and tested in the no-task condition.

The original data set was characterized for sensitivity to directed attention during the initial periods of the trials, specifically the instruction, standard and salient texture periods (Fig. 1), by comparing discharge rates during the tactile and visual discrimination tasks (Meftah et al. 2002). This subset of cells is representative of the original data set in that a low proportion of S1 cells (5/53) as compared with S2 cells (28/50) was attention-sensitive in the salient texture period, and the distributions of attention-sensitive cells did not differ from the original samples (² tests, P > 0.20). The validity of comparisons across S1 and S2 is supported by the fact that both areas were sampled in each monkey. In addition, performance levels during the task trials acquired during sampling from each region were similar (error rates of, respectively, 4.2 ± 0.4 and 3.5 ± 0.4%; P = 0.20). Two representative examples, both sensitive to surface texture, are shown in Fig. 2. The S1 cell (Fig. 2A) was classified as non-attention sensitive (tactile task, black line; visual task, red line). Its firing rate showed an abrupt increase after the start of surface scanning (standard). Discharge rates to the standard 2 mm surface then adapted to lower levels. When the digits later encountered the modified surface, there was a second increase in firing rate. Although there was a trend for discharge rates to be higher in the tactile task during the salient texture period, the difference was not significant (P = 0.07). The S2 cell (Fig. 2B) was, in contrast, attention-sensitive in all three of the early periods of the trials, firing rates being higher when attention was directed to the tactile modality as compared with visual.

This report concentrates on single unit data collected during the salient texture period (1), the postreward period (2), and the static period (3). Analyses for period 1 were focused on trials from the tactile discrimination task, so attention was directed to the tactile modality, and surface texture was salient. During periods 2 and 3, attention was directed to the liquid reward, and surface texture was no longer salient.

Single-cell examples

S1 texture-sensitive cells were characterized by signaling surface texture independent of the direction of attention. Figure 3 shows a typical example, recorded in area 2. During the salient texture period (Fig. 3A, rasters and filled histograms), a prominent increase in discharge was observed when the digits first encountered the modified surface textures (3.7 and 4.7 mm) as compared with the discharge evoked by the standard surface (2 mm). Firing rates were lower in the postreward period but continued to reflect the underlying surface texture. No evidence of texture sensitivity was found in the static period in this and most other cells. Inspection of the summary plot in Fig. 3B (thick lines) shows that there was a monotonic relation between mean discharge rate and SP during both periods 1 and 2 of the task (P < 0.0005 and P = 0.023, respectively). Discharge rates were, however, significantly lower during period 2/task, as compared with period 1/task (paired t-test, P < 0.0005), reflecting some adaptation of discharge rates and/or the direction of attention (respectively, to the reward or the tactile stimulus).

View larger version (30K): [in this window] [in a new window]   FIG. 3. Representative S1 texture-sensitive neuron (same cell as shown in Fig. 2A). A: rasters and filled histograms show discharge during the task recordings as a function of the 3 SPs and 3 analysis periods. Note the change in time scale for period 1 vs. periods 2 and 3. For the salient texture period (1), the responses to the modified surfaces (3.7 and 4.7 mm SP, right, period 1b in Figs. 1 and 2) were taken from the same trials shown on the left (2 mm SP, period 1a in Figs. 1 and 2). For the postreward and static periods, data were pooled from the tactile and visual tasks, with the exception of the 2 mm SP responses, which were taken from trials in the visual task with no texture change. Cell discharge frequency increased with increasing SP during both dynamic periods (1 and 2) but not the static period (3). Sensitivity to SP was preserved in the no-task condition (open histograms, rasters not shown), but firing rates were lower in period 1. B: summary plot showing mean firing rates as a function of SP and analysis period for both task conditions (task, thick lines; no-task, thin lines). The data from period 3 were collapsed across the 3 SPs. Task recordings showed significant enhancement in period 1 (task vs. no-task) but not periods 2 or 3. C: mean vertical contact force during periods 1 and 2 for the data shown in B.

S2 texture-sensitive neurons, in contrast, were very sensitive to the salience of the stimulus, showing enhancement during the salient texture period and suppression after the discrimination response was made. The cell illustrated in Fig. 4 is a typical example. During the salient texture period of the task, this cell showed a pattern of increased discharge with increased SP (A and B, P < 0.0005). In the same trials, firing rates declined precipitously in the postreward period when surface texture was no longer salient, and sensitivity to texture disappeared (P = 0.262). During the no-task condition, weak texture sensitivity was apparent in both dynamic periods (P < 0.01). Examination of the summary plot (Fig. 4B) suggests that there were two opposing processes during the task recordings: enhancement of salient inputs followed by suppression of neuronal sensitivity to tactile inputs after the motor response, at a time when attention was directed to the juice reward (task vs. no-task, P < 0.0005). The suppression continued into the static period because discharge was also significantly lower than that seen in the no-task condition (P < 0.0005).

Vertical contact force on the textured surfaces was measured in each analysis period. Inspection of the force traces from the recordings (Fig. 2) and the summary plots (Figs. 3C and 4C) indicates that there was no systematic change in contact force across the three analysis intervals. In particular, there was no evidence that the dramatic decline in discharge frequency in the postreward period for the S2 cell was accompanied by a change in contact force (Fig. 4C). There was, on the other hand, a trend for slightly lower contact forces during the no-task recordings as compared with the task condition (e.g., Fig. 4C but not 3C), but any such changes were small, often overlapping the values recorded during the task recordings.

Population analysis: S1

Overall, 74% (Table 1) of the S1 cells preserved their texture-sensitivity in the postreward period (task), including 24 of 32 cells showing increased discharge with increased SP (T+) and 2 of 3 cells showing the opposite response pattern (T–, decreased discharge with increased SP). Figure 5 (left) shows the pooled task data for the 32 T+ cells. Discharge frequency varied as a function of both the analysis period and SP (repeated-measures ANOVA, P < 0.0005). Post hoc comparisons showed that discharge frequency was higher in the salient texture period as compared with the postreward period (P < 0.0005). Discharge in the latter period was, in turn, higher than that in the static period (P < 0.0005). These differences were independent of variations in vertical contact force (P = 0.274).

View larger version (14K): [in this window] [in a new window]   FIG. 5. Population analysis of S1 and S2 in the task condition. Mean firing rate of the T+ cells (increased discharge with increased SP) is plotted as a function of SP and analysis period. Neuronal responsiveness to tactile stimuli varied as a function of the analysis period, being highest when the input was salient (salient texture), and lower when the input was not behaviorally relevant (postreward, static). S1 cells preserved their texture sensitivity even when the input was not relevant. This contrasted with S2 neurons that showed a large drop in firing rate in the postreward period along with reduced dependence on SP. In neither region did discharge covary with surface texture during the static period.

View larger version (26K): [in this window] [in a new window]   FIG. 6. Effect of task condition (task, thick lines; no-task, thin lines) on the activity of S1 and S2 cells. For comparison, data from the visual task (period 1: SPs of 2 and 4.7 mm; red symbols) are also shown. A: task condition significantly modulated the responsiveness of S1 and S2 cells that were categorized as texture-related in the task. B: non-texture-related cells showed a small increase in discharge frequency during the salient texture period (vs. postreward period), but no significant change in firing rate with task condition. T, tactile; V, visual.

Population analysis: S2

Only 47% of S2 cells preserved their texture-sensitivity in the postreward period (Table 1), including 14/33 T+ cells and 3/3 T– cells. The pooled results from the T+ cells (task) are plotted in Fig. 5 (right). Discharge frequency varied as a function of the analysis period and SP (P < 0.0005). Post hoc comparisons confirmed that discharge in the postreward period was significantly lower than that in the salient texture period, and higher than in the static period (P < 0.0005). As also found for S1, vertical contact force did not vary systematically across the three periods and two task conditions (P = 0.922).

Nineteen of 36 texture-sensitive S2 cells were tested in the no-task condition (18 T+ and 1 T–). Figure 6A (right) summarizes the results in the task and no-task conditions for the T+ cells. The pattern of response modulation mirrors that shown for the individual cell example in all respects. Discharge frequency was enhanced in the salient texture period (task) for all three surface textures as compared with all other analysis intervals (task and no-task). In the later postreward period, discharge was decreased in comparison to the no-task recordings. This suppression continued on into the static period.

A repeated-measures ANOVA revealed that there were significant changes in discharge as a function of the two task conditions (P = 0.012), the two dynamic periods (P < 0.0005) and three SPs (P < 0.0005). All of the interaction terms were also significant (P < 0.0005), reflecting the complex pattern of modulation. Post hoc contrast analyses demonstrated that discharge frequency during period 1 was significantly higher during the task (P = 0.001) as compared with the no-task condition. The opposite result was obtained in period 2: discharge during the postreward period/task was significantly lower than in the no-task condition (P = 0.001). When vertical contact force was included as a covariate in an ANOVA, force was, however, a significant factor (P = 0.007). The analysis was then repeated after normalizing cell discharge (see METHODS). Contact force was no longer significant (P = 0.31), suggesting that differences in cell discharge rates contributed to the result.

For comparison, the results in 14 non-texture-related cells recorded during both the task and the no-task conditions are plotted below (Fig. 6B, right). The results were similar to those observed for S1 non-texture-related cells. There was no significant change in discharge with task condition (ANOVA, P = 0.41), but there was a significant change across the two dynamic stimulation periods (P = 0.016). This was explained by higher discharge during period 1/task as compared with period 2/task (post hoc contrast: P = 0.021). The relative magnitude of the increase, 55%, was much less than that seen for texture-sensitive S2 cells, 235%, in part because there was no evidence for a postreward suppression of discharge (period 2/task vs. period 2/no-task, P = 0.113).

Comparison of S1 and S2

Taking discharge frequency in the no-task condition as the reference value, the pooled data shown in Fig. 6 indicate that there was an average 24% enhancement of discharge in S1 texture-related cells during the salient texture period, attributed to directed attention. A similar, but nonsignificant, increase was observed in the non-texture S1 and S2 cells, respectively, 26 and 29%, during the same period. A threefold larger (82%) enhancement in discharge rates was observed in the S2 texture cells during the salient texture period, followed by a suppression of discharge (31%) during the postreward period. This contrasted with the results obtained in S1 texture and non-texture S1 and S2 cells where discharge rates in the postreward period were similar to that seen in the no-task condition.

To determine how neuronal sensitivity to tactile stimuli was differentially modified in S1 and S2, we fit linear regression curves, firing rate versus SP, to the individual cell data summarized in Fig. 6 for each analysis period. Separate curves were fit to the data acquired from each recording condition, task and no-task. In this analysis, the slope provided an estimate of the gain of the stimulus-response function (multiplicative factor), while the intercept described the offset of the function (additive factor) (Andersen et al. 1985; Mitchell and Silver 2003; Salinas and Thier 2000). The slopes of the resulting regression curves, task versus no-task, are plotted in Fig. 7A (texture-related cells only).

View larger version (26K): [in this window] [in a new window]   FIG. 7. A and B: comparison of S1 and S2 texture-related cells. A: distribution of slopes for linear regressions (firing rate vs. SP) in the task as compared with the no-task condition. Shown here are data from periods 1 (blue: salient texture vs. early), 2 (green: postreward vs. late), and 3 (black: static). The corresponding Gaussian bivariate confidence ellipse (1 SE) is shown for each period. Cells that showed a significant correlation in both conditions are shown with filled symbols; others are shown with open symbols. The diagonal line represents equality. Task condition had little effect on the slopes of S1 texture-related cells. In contrast, S2 texture-related cells were strongly influenced by task condition and the analysis interval. B: modulation index, MI = (task – no-task)/(task + no-task), as a function of SP (period 1), calculated from predicted and experimental data (respectively, interrupted and solid lines). The experimental data fit an additive increase in discharge for S1 (left) and a multiplicative increase in discharge for S2 (right). The predicted curves used the mean discharge of the texture-related cells during the no-task condition as the baseline.

In contrast, the slopes of the regressions for texture-related S2 cells were systematically modified by task condition across the three analysis intervals. Inspection of Fig. 7A (right) shows that the length and orientation of the major axis of the confidence ellipses differed (linear discriminant analysis, P < 0.0005). For period 1, all but one data point (blue) was located above the equality diagonal, i.e., slopes were significantly higher (2 times) during the salient texture period as compared with the no-task condition (task, 10.6 ± 1.2; no-task, 4.9 ± 1.0; P = 0.002). In contrast, the majority of period 2 (14/18, green) and period 3 (12/18, black) data points were located below the diagonal, and the slopes for period 2/task were significantly lower than for the No-task condition (respectively, 1.6 ± 0.5 and 3.3 ± 0.4, P = 0.006). No differences in intercepts were observed. Thus directed attention during the salient texture period increased the slopes of the S2 stimulus-response curves. In other words, response gain was increased. The net effect was to increase discharge rates, amplifying the relative difference in discharge frequency elicited by the modified textures (3.7 and 4.7 mm) as compared with the standard 2 mm texture. When attention was subsequently directed toward the reward, slopes were dramatically lower than for the no-task condition, i.e., response gain was decreased. Consistent with this latter observation, significantly fewer S2 than S1 cells were texture sensitive after the discrimination response was made (Table 1; ² test, P = 0.020). Non-texture-related S2 cells (not shown), on the other hand, showed no change in slope across the two dynamic periods (discriminant analysis, P = 0.46), consistent with the context-dependent effects in S2 being focused on texture-sensitive cells.

The results of the linear regression analyses were confirmed in a second analysis that made no assumption as regards the nature of the relation between discharge frequency and SP for texture-related cells. This approach was justified because some cells (e.g., Fig. 4) showed a nongraded increase in discharge for the modified surfaces (see also Jiang et al. 1997). The magnitude of modulation for each SP, during both dynamic analysis periods, was calculated using the modulation index, MI = (task – no-task)/(task + no-task). The results from period 1 are plotted in Fig. 7B, in relation to the distribution of theoretical data based either on an additive (left) or multiplicative (right) model. These models were chosen to reflect the experimental observations (see preceding text). Observed mean discharge rates in the no-task condition served as the baseline. For texture-related cells, the additive model predicted a decrease in MI as SP increased. The S1 data fit this prediction well, close to the prediction based on a mean increase in discharge frequency of 12.5 imp/s (Fig. 7B, left), as compared with the observed increase of 11 imp/s. An ANOVA confirmed this impression: MI varied across the three SPs (P = 0.016) and two analysis intervals (P = 0.033), with MI being close to 0 for period 2. For S2, in contrast, MI varied across the two analysis intervals (means of +0.278 and –0.224 for periods 1 and 2) but not across the three SPs (P = 0.211). The latter observation was consistent with the multiplicative model. The period 1 data were closest to the prediction associated with a 1.75 times increase in slope (observed, 2.16 times) during period 1 (Fig. 7B, right). MI for period 2 (S2, not shown) was negative, corresponding to a multiplicative factor of 0.6 times (observed change in slope, 0.5 times). The small differences relative to the measured changes in slope likely represent limitations in the linear model employed.

No-task condition

One assumption in this study was that attention during the no-task recordings was directed away from the tactile stimuli and toward the occasional drops of juice. To test this assumption, we measured the firing rates in the visual discrimination task trials, during the early interval (periods 1a and 1b, Fig. 1) and compared the results to those obtained during the same interval in the corresponding texture discrimination task trials, and the no-task recordings. The analyses were restricted to rewarded visual trials, and so attention was directed toward the visual modality and away from the tactile modality. Given the suppression seen in S2 during the postreward period, the data from the visual task were necessarily restricted to trials in which the modified texture preceded the discrimination response (and reward) and so preceded the salient change in visual intensity. Consequently, the analyses were restricted to trials with two SPs, the 2 mm standard texture and the 4.7 mm modified surface.

For the texture-sensitive cells (red symbols, Fig. 6A), firing rates during period 1 varied significantly across the three conditions (tactile task, visual task, no-task: repeated-measures ANOVA, P = 0.002 for S1; P < 0.0005 for S2). Post hoc comparisons demonstrated that firing rates in the tactile task were higher than for the visual task (S1, P = 0.015; S2, P < 0.0005), consistent with directed attention enhancing firing rates in both areas during the salient texture period. There was no difference across the visual and no-task recordings for either S2 texture- or S1 and S2 non-texture-related cells. These observations support our assumption that attention was directed away from the tactile modality in the no-task condition. For S1, in contrast, discharge rates during the visual task were higher than in the no-task recordings (P = 0.048). Similar results were obtained when we compared the MI (period 1) to the attention modulation index, AMI = (tactile – visual)/(tactile + visual), calculated for the same analysis interval (Meftah et al. 2002). MI was significantly higher than AMI for texture-related cells in S1 (P = 0.007) but not S2 (P = 0.092). Non-texture-related cells in S1 and S2 showed no difference (MI = AMI). The significant result for S1 texture-related cells cannot be explained by systematic changes in recording conditions because discharge rates in period 2 were identical for the task and no-task recordings. The visual trials differed from the no-task trials in two ways, intention to respond and motivation, and these factors may have contributed to the observed difference. We suggest that the significant difference for S1 texture-related cells may reflect the intention to move (Nelson 1988) because S1, but not S2, receives input from primary motor cortex (Jones and Wise 1977).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
The results show that there are fundamental differences in the way in which attention or behavioral context modulates the discharge of S1 and S2 neurons. S1 texture- and non-texture-related cells were characterized by enhanced responses (25%) to scanned textures during the salient texture period, but there was no significant change in response gain. In contrast, S2 texture-related cells, but not non-texture-related cells, showed a threefold greater enhancement of discharge with directed attention along with increased response gain. When attention was subsequently directed to the reward, only S2 texture-related cells showed a marked suppression of their discharge and decreased response gain. Altogether these results are consistent with the existence of at least two independent attentional mechanisms, one generalized and the other focused specifically on S2 texture-related cells.

General consideration

One underlying assumption in this study was that the texture discrimination task itself was essentially a spatial discrimination task based on the physical distance between the raised dots and signaled by changes in the mean firing rate of S1 and S2 cortical neurons as a function of the spatial parameters of the stimulation. Another interpretation is, however, that the monkeys may have based their decisions on measuring a change in the temporal interval between successive rows of raised dots. While we cannot rule out this latter possibility, as scanning speed was fixed, we have demonstrated previously that S1 neurons show graded changes in mean firing rate over the range used here (Jiang et al. 1997) (note: S2 neurons showed nongraded changes in the same study; this result was not reproduced here, probably because task requirements and the analysis window differed). Moreover, a substantial proportion of S1 cells signal surface texture independent of changes in scanning speed (Tremblay et al. 1996), an observation that we have recently extended to a wide range of SPs (1.5–8.5 mm) (Bourgeon et al. 2005). Consistent with the latter finding, we have also shown that roughness perception in humans is independent of scanning speed (Meftah et al. 2000). Together, we believe that the spatial, and not the temporal, parameters of the tactile stimuli were salient for task performance, but further experiments are needed to support this suggestion.

S1 cells

S1 texture-related cells showed one major change across the test conditions, increased responses to scanned surfaces during the salient texture period as compared with the postreward period and to the no-task testing. These effects were independent of any associated changes in contact force. Along with our previous results (Meftah et al. 2002) showing that directed attention has little or no effect on discharge during earlier epochs of the same trials (instruction and standard stimuli periods), it appears that directed attention has very focused effects in S1, limited to the interval containing the salient change in texture. Although relatively few texture-related S1 cells showed significant changes in discharge with directed attention (26% in Meftah et al. 2002), the present results now show that firing rates for the sample of S1 cells during the salient texture period were significantly increased during the tactile task, both in relation to the no-task recordings and also to the visual task recordings. These results could not be explained by the change in the method used to measure firing rate during the salient texture period (adjusting the window on a trial-by-trial basis here vs. a standard window starting when the texture change first entered the aperture) as the mean change in firing rates elicited by the texture change (texture-sensitive cells) were, in agreement with our previous results, similar in the tactile and visual tasks (respectively, 34 ± 3 and 31 ± 4 imp/s, P = 0.22), as expected if response gain in S1 was not altered by the direction of attention.

Our finding of higher discharge rates in the task versus no-task condition is consistent with previous studies using other types of tactile stimuli, including probe movement across the skin (Romo et al. 1996) and flutter vibration (Salinas et al. 2000). During tactile trials in our paradigm, attention was directed toward the surface texture during the initial presentation of the standard surface texture (2 mm SP) and during the salient texture period. In the subsequent postreward period, attention was no longer controlled but was likely directed toward the juice reward triggered by the correct response.

Our previous report was unable to determine if attention had enhanced the salient inputs or suppressed irrelevant inputs. This study has now shown that in the same trials, discharge frequency in the postreward period dropped down to levels comparable to that seen in the no-task condition when attention was no longer directed to the tactile stimulus. We interpret the results as providing evidence that directed attention selectively enhanced S1 neuronal responsiveness to relevant tactile stimuli during the salient texture period. Although the effect appeared to be generalized because texture- and non-texture-sensitive S1 cells showed qualitatively similar increases during the salient texture period (24 vs. 26%), the absolute increase in discharge was greater for texture- as compared with non-texture-related cells (11 vs. 4.1 imp/s). The magnitude of the change was, on the other hand, small in comparison to S2 texture-related cells (82%, 20.3 imp/s).

We suggest that the underlying mechanism is an additive increase in neuronal responsiveness. This suggestion is supported by the results of both the linear regression and MI analyses (Fig. 7). We further suggest that the response enhancement in S1 cells was restricted in time, approximately to the period that spanned the times that the texture change might occur. The onset of the modulation was not directly determined but, as previously reported, discharge rates of S1 cells during the initial 700-ms presentation of the standard tactile and visual stimuli were rarely modified by the direction of attention, with only 3/102 S1 cells showed attention-related changes in discharge during both the salient texture period and the preceding standard stimulus period. The latter period preceded the texture change by a minimum 0.5 s (early texture change). Thus modulation began no earlier than 500 ms prior to the texture change. Response enhancement was not evident in the postreward period (final 400 ms of the trial), a period that began within 200 ms of the successful discrimination response (trials with late texture change). This suggests that the response enhancement likely ended shortly after, or coincident with, the motor response that signaled discrimination of the change in texture.

S2 cells

In our previous report (Meftah et al. 2002), we found that S2 cutaneous cells showed more frequent, larger and more complex attentional modulation than did S1 cells. Moreover, the effects were not focused on the period with the salient change in texture but frequently encompassed the earlier epochs of the same trials. The present results extend these observations, showing that task context modulated S2 discharge in later epochs of the trials and that S2 cells were differentially modulated across the different test conditions as a function of whether or not they were categorized as texture sensitive. Non-texture-sensitive S2 cells showed changes that resembled those seen in S1 (increase of 4.4 vs. 4.1 imp/s for S1 non texture-related cells). The modest enhancement in the salient texture period may have reflected the initial processing in S1 or alternately the existence of a more generalized additive attentional mechanism at both levels. Texture-sensitive S2 cells (all T+) also showed an enhancement of responses in the salient texture period, but the magnitude of the increase was threefold greater. While this increase likely reflected, in part, the additive mechanism invoked in the preceding text, detailed analyses (regressions, MI) suggest that a second mechanism was involved, an increase in response gain. The advantage of such a multiplicative mechanism is that differences between the modified and standard surface textures would be enhanced and so contribute to the performance of this texture discrimination task consistent with observations from psychophysical experiments in humans (Zompa and Chapman 1995). No gain change was seen in S1, suggesting that the multiplicative mechanism was only operational in S2. Finally, the gain change could be specifically attributed to attention because, as reported previously, the mean change in discharge frequency seen during the salient texture period (texture-sensitive S2 cells) was significantly higher during the tactile task as compared with the visual task (Meftah et al. 2002), an observation that was confirmed here using a modified analysis method (respectively, 28 ± 3 and 18 ± 2 imp/s, P = 0.001).

S2 texture-related cells were also characterized by showing a pronounced suppression of their discharge in the postreward period along with significantly lower response gains (vs. no-task). As a consequence, S2 cells were significantly less likely to signal texture differences than S1 cells during the postreward period. Although it seems most likely that attention was responsible for the dramatic decrease, two other factors may have contributed. The absence of any postreward suppression in the non-texture-sensitive cells argues against the possibility that tactile responsiveness was suppressed or gated after the discrimination response (lever release with the contralateral, nonstimulated hand). Although we cannot exclude the possibility that the result reflects the existence of a very targeted movement-related suppression of tactile inputs limited to texture-sensitive S2 cells, available evidence indicates that movement-related gating is best characterized as being nonspecific (Chapman 1994) and spatially restricted (Williams et al. 1998) so that only inputs from the moving limb are suppressed. A second potential contributing factor was reward expectancy. This was high during the salient texture period (prereward) and zero in the postreward period (only rewarded trials included in these analyses). For the no-task recordings, reward expectancy was also low because there was no explicit stimulus associated with the quasi-random drops of juice given at intervals of 4–7 s (Fiorello et al. 2003). Thus the added decrease in the postreward period (vs. no-task) cannot be explained by differences in reward expectancy leaving attention as the most likely explanation. The question then arises as to whether this suppression (multiplicative as response gain was significantly decreased) was obligatorily linked to the preceding response enhancement or was perhaps an independent process. We favor the former suggestion given that slope changed significantly with task condition across both intervals. Moreover, inspection of the data from each cell confirmed that all but one cell showed clear evidence for both initial enhancement followed by response suppression.

The presence of postreward suppression in S2, but not S1, suggests that each area plays a different role in the analysis of cutaneous inputs. S2 signals information regarding SP when it is relevant to ongoing behavioral demands, whereas S1 signals SP independent of context or other behavioral factors. Our observations fit well with Itti and Koch (2001)’s prediction, based on a consideration of computational models of visual attention, that an inhibitory mechanism is required to allow neuronal resources to actively disengage from attended inputs, ensuring that attention can be shifted elsewhere. In this light, the postreward suppression may reflect a resetting of neuronal activity in preparation for the next trial. In addition, and as pointed out by Itti and Koch, the suppressed discharge may provide a neuronal correlate for the psychophysical phenomenon of "inhibition of return" whereby sensory processing at recently attended locations may be slowed when the stimulus—visual, auditory or tactile—is repeated with a short interstimulus interval (Klein 2000; Spence et al. 2000).

Sensitivity to static surface texture

The discharge of few cells in either area reflected the underlying static surface texture, regardless of the task condition. Only 14% of S1 and 8% of S2 cells continued to signal texture differences under the static condition. The physical parameters of the raised dot surfaces were clearly within the range that can be discriminated by humans using static touch (Van Boven and Johnson 1994), and so it was reasonable to expect that cell discharge would reflect SP. The low proportion of static texture-sensitive cells is likely explained by the high proportion of cells with a receptive field encompassing more than one digit tip (28/35 in S1, 35/36 in S2). For these cells, asymmetric stimulation of the receptive fields on the adjacent digits in contact with the surface may have obscured any covariance of discharge with SP. Consistent with this interpretation, three of the five S1 cells sensitive to static SP had a receptive field restricted to only one digit tip. Thus we likely underestimated the proportion of static texture-sensitive cells.

Independent attentional controls in S1 and S2

We suggest that the present results are consistent with the existence of independent attentional controls in S1 and S2. The alternate hypothesis that the effects seen in S1 reflect top-down controls from S2 cannot be discounted entirely, but this seems unlikely. First, although both areas showed enhanced discharge during the salient texture period, the basic mechanisms differed—an additive increase in responsiveness in S1 during the salient texture period versus a multiplicative increase in response gain in S2. Second, S1 cells showed no evidence for the postreward suppression that characterized S2 texture-sensitive cells.

The present observations are consistent with a two-stage modulation of somatosensory cortical discharge—an initial stage (S1) with increased tactile responsiveness during the salient texture period, followed by enhanced response gain at a later stage (S2) presumably contributing to improving feature extraction. These latter effects were, moreover, focused specifically on the task-relevant signals as response gain was only increased in S2 texture-related cells. We suggest that the effects seen in S2 likely reflect the sum of the attentional influences exerted in S1 plus further controls at the level of S2. Such an organization would explain the modest enhancement of discharge of non texture-sensitive S2 cells during the salient texture period. Although the neuronal mechanisms remain to be determined, at the very least it is unlikely that simple convergence from S1 cells onto S2 could explain the gain changes in S2, one process being linear and the other nonlinear.

Returning to the questions addressed by this study, we conclude that sensitivity to SP is differentially modified by directed attention in S1 and S2, directed attention produces a net enhancement of responses to attended tactile stimuli in both areas, and attentional effects are generalized in S1 but focused in S2 on texture-sensitive cells. Taken together, the present results are consistent with S1 and S2 playing different roles in the analysis of tactile stimuli. Our results in S1 are consistent with previous studies showing that S1 cells provide a faithful representation of stimulus characteristics across a range of behavioral conditions (Hernández et al. 2000; Jiang et al. 1997; Salinas et al. 2000). The more complex effects seen in S2 are, on the other hand, consistent with suggestions that S2 discharge reflects more ongoing behavioral demands associated with a perceptual task (Jiang et al. 1997; Romo et al. 2002), and this in a manner consistent with S2 being located at a hierarchically higher level than S1 in the processing of tactile stimuli (Pons et al. 1992).

Comparison with studies of visual attention

In more general terms, several parallels can be drawn with findings from studies of directed attention in the visual system. The proposed additive enhancement of neuronal responsiveness in S1 and S2 cortex is reminiscent of an additive increase in baseline discharge observed by Luck et al. (1997) in V2 and V4 but not V1. But it seems unlikely that the additive increase seen here can be explained by baseline changes as we found little evidence for baseline shifts in S1 with directed attention (3% of cells), although shifts were more frequent in S2 (20%) (Meftah et al. 2002). This observation highlights a potential difference in attentional mechanisms for somatosensory as compared with visual stimuli. In contrast, the multiplicative enhancement seen in S2 cortex is consistent with reports of attention-related multiplicative scaling of orientation tuning curves in V4 (e.g., McAdams and Maunsell 1999; Spitzer et al. 1988) and direction tuning curves in the middle temporal (MT) visual area (Treue and Martinez Trujillo 1999). Controversy exists as to whether these changes are accompanied by increased response selectivity, specifically narrowing of tuning curves. In this regard, it would be interesting to extend our study to include a larger range of modified surface textures to determine whether, as reported in V4, the stimulus-response curve might be shifted with directed attention (Reynolds et al. 2000). Such a finding would be consistent with Reynolds et al.’s suggestion that (visual) stimulus intensity and features (e.g., orientation, direction) are processed differently. Finally, we know of no evidence from studies of visual attention of the striking postreward depression of neuronal responsiveness to tactile stimuli seen here in S2. This likely reflects differences in the experimental paradigms as visual attention studies rarely employ long-duration stimuli. Given the potential functional importance of this mechanism for actively disengaging neurons from recently attended stimuli, it would be interesting to determine whether such a mechanism also exists within the visual system.

	GRANTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS ACKNOWLEDGMENTS REFERENCES

 
The research was supported by grants from the Canadian Institutes of Health Research and the Université de Montréal. E.-M. Meftah was supported by bursaries from the Groupe de Recherche sur le Système Nerveux Central (research center supported by the Fonds Pour la Formation de Chercheurs et l’aide à la Recherche) and the Centre de Recherche en Sciences Neurologiques (Université de Montréal).

	ACKNOWLEDGMENTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION GRANTS