Behavioral/Systems/Cognitive Neuronal Basis of Covert Spatial Attention in the Frontal Eye Field Kirk G. Thompson, Keri L. Biscoe, and Takashi R. Sato Laboratory of Sensorimotor Research, National Eye Institute, National Institutes of Health, Bethesda, Maryland 20892 The influential ���premotor theory of attention��� proposes that developing oculomotor commands mediate covert visual spatial attention. A likely source of this attentional bias is the frontal eye field (FEF), an area of the frontal cortex involved in converting visual information into saccade commands. We investigated the link between FEF activity and covert spatial attention by recording from FEF visual and saccade-related neurons in monkeys performing covert visual search tasks without eye movements. Here we show that the source of attention signals in the FEF is enhanced activity of visually responsive neurons. At the time attention is allocated to the visual search target, nonvisually responsive saccade-related movement neurons are inhibited. Therefore, in the FEF, spatial attention signals are independent of explicit saccade command signals. We propose that spatially selective activity in FEF visually responsive neurons corre- sponds to the mental spotlight of attention via modulation of ongoing visual processing. Key words: vision saccade attention monkey physiology premotor Introduction Humans and monkeys are able to select and acquire visual infor- mation preferentially within a locus of peripheral vision without shifting gaze (Posner, 1980 Kinchla, 1992 Egeth and Yantis, 1997). This ability, known as covert spatial attention, often is compared metaphorically with a mental spotlight that illumi- nates a selected area or object for enhanced processing. Behav- ioral studies have shown that covert spatial attention and overt eye movements are closely linked (Hoffman and Subramaniam, 1995 Kowler et al., 1995 Sheliga et al., 1995a,b Deubel and Schneider, 1996) and support the premotor theory of attention, which proposes that covert attention arises from latent eye move- ment commands even when eye movements are not made (Riz- zolatti et al., 1987 Sheliga et al., 1995a Moore et al., 2003). Ad- ditional support for this theory comes from studies of the frontal eye field (FEF), an area in the frontal cortex that, in addition to generating saccade commands (Bruce and Goldberg, 1985 Bruce et al., 1985 Hanes and Schall, 1996 Tehovnik et al., 2000), plays a central role in the allocation of spatial attention in both humans (Corbetta et al., 1998 Beauchamp et al., 2001 Corbetta and Shul- man, 2002 Grosbras and Paus, 2002 Muggleton et al., 2003 Kincade et al., 2005) and monkeys (Moore et al., 2003 Moore and Fallah, 2004). This evidence has led to the hypothesis that FEF saccade-related movement neurons mediate covert attention by modulating the gain of neurons in extrastriate visual cortex (Hamker, 2005). Although many studies are consistent with the premotor theory of attention, the fact that the FEF plays a role in both covert attention and eye movements does not mean neces- sarily that covert attention and eye movements originate from the same source they could be mediated by different processes (Klein and Pontefract, 1994). In monkeys performing visual search tasks traditionally used to study visual attention, the activity of FEF neurons evolves to identify the target of a search array before a saccade is made (Schall and Hanes, 1993 Schall et al., 1995b Thompson et al., 1996, 2005 Bichot et al., 2001a,b Sato et al., 2001 Sato and Schall, 2003). This selection process does not depend on saccade production (Thompson et al., 1997 Murthy et al., 2001 Sato et al., 2003). However, in all of these studies, the saccades were a prominent component of the either the task or the monkeys��� training. Therefore, it could be argued that the selection process could reflect some component of saccade planning. In this study, we recorded from single neurons in the FEF while monkeys performed a pop-out visual search task requiring a manual response in which there was clear evidence of absence of saccade planning. First, we tested the hypothesis that there is activity in the FEF that corresponds to the locus of attention during visual search that cannot be attributed to previous train- ing or to saccade production. Second, we tested the specific pre- diction that, when attention shifts covertly to a target in the visual field, motor activity for a saccade toward the locus of attention also should be present. We found that a spatially selective signal that could correspond to the spotlight of attention was present in the activity of most visually responsive FEF neurons, but the ac- tivity of movement neurons was suppressed. Materials and Methods Data collection. Two experimentally naive male monkeys (Macaca mu- latta), weighing 8 kg (monkey S) and 6.5 kg (monkey C), were prepared Received Feb. 23, 2005 revised Aug. 17, 2005 accepted Aug. 18, 2005. This work was supported by the Intramural Research Program of the National Institutes of Health���National Eye Institute. We thank N. Bichot, J. Schall, and our colleagues in the Laboratory of Sensorimotor Research (J. Cavanaugh, N. Port, B. Sheliga, M. Sommer, R. Wurtz, and H. Zhou) for helpful discussions and valuable comments. Correspondence should be addressed to Kirk G. Thompson, Laboratory of Sensorimotor Research, National Eye Institute, National Institutes of Health, Building 49, Room 2A50, Bethesda, MD 20892. E-mail: kgt@lsr.nei.nih.gov. T. R. Sato���s present address: Cold Spring Harbor Laboratory, 1 Bungtown Road, Cold Spring Harbor, NY 11724. DOI:10.1523/JNEUROSCI.0741-05.2005 Copyright �� 2005 Society for Neuroscience 0270-6474/05/259479-09$15.00/0 The Journal of Neuroscience, October 12, 2005 ��� 25(41):9479 ���9487 ��� 9479

for electrophysiological recordings. All surgical and experimental proto- cols were approved by the National Eye Institute Animal Care and Use Committee and complied with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Sterile surgery was performed under ketamine and isoflurane anesthesia to place a head-holding device, a plastic recording chamber over the left FEF, and a scleral search coil. The FEF was localized within the recording chamber by using low current microstimulation ( 50 A) to evoke saccades and by the presence of saccade-related movement neurons (Bruce and Goldberg, 1985). Re- cording sites were confirmed to be in the rostral bank of the arcuate sulcus histologically in monkey S and by magnetic resonance imaging in monkey C. Visual stimulation and behavioral control were done by a computer running the real-time experimentation data acquisition system (REX) (Hays et al., 1982). Visual stimuli were presented on a computer monitor (26 21 cm 1024 768 pixel resolution 85 Hz frame rate) viewed at a distance of 57 cm. Action potential waveforms were recorded with tung- sten microelectrodes, digitized, and saved by using a computer-based data acquisition system (Plexon, Dallas, TX). Often two or three units were recorded simultaneously. Off-line spike sorting separated single units on the basis of size and shape of the spike waveforms. Analog eye position and lever position signals were digitized and sampled at 1 kHz. Behavioral training and tasks. The monkeys used in this experiment had no previous experience in performing behavioral tasks. Monkeys were seated in a primate chair with the head fixed. Using operant condi- tioning with positive reinforcement, we trained the monkeys to perform a memory-guided saccade task and a covert visual search task. The two tasks were run in separate blocks of trials. The memory-guided saccade task was used to distinguish visual from movement activity for cell classification and to map the spatial extent of the response field of each neuron (Bruce and Goldberg, 1985) (see Fig. 1a). After the monkey fixated on a 0.3�� diameter gray spot on a black background for 400���800 ms, an identical spot was flashed for 50 ms at a peripheral location. The monkeys were required to maintain fixation on the central spot for a random interval ranging from 800 to 1400 ms. After the central spot disappeared, the monkeys were rewarded for making a saccade to the remembered location of the target. Once gaze shifted, the target reappeared to provide feedback and a fixation target for the monkeys. Covert visual search tasks were used to examine neural activity during visual search without saccades (see Fig. 1b). A lever that could be turned left or right of vertical was attached to the front of the chair within easy reach of the monkey. When no force was applied to the lever, a spring automatically returned it to the vertical position. Although the monkeys were free to use either hand to turn the lever, monkey S was exclusively left-handed and monkey C was exclusively right-handed. The location and identity variations (see Fig. 1b) of this task had the same temporal structure. After the monkey grasped the lever and posi- tioned it within 10�� of vertical, a small (0.3��) central yellow fixation cross appeared on a black background. The different fixation stimulus was used to help distinguish this task from the memory-guided saccade task. In this task, the monkeys were required to maintain fixation on the central stimulus until the reward. After the monkeys fixated on the cen- tral cross for a random interval (400���800 ms), a target was presented randomly at one of six or eight isoeccentric locations spaced equally around the fixation cross. The remaining locations were occupied by distractors. Each of the stimuli subtended 1.5�� of visual angle, and the eccentricity of the stimuli was adjusted so that at least one of the stimulus locations was inside the receptive field of the neuron. The monkeys were rewarded for making the correct lever turn ( 15�� from vertical) within 2 s after search array presentation in practice, the monkeys nearly always turned the lever to the limit of 35�� from vertical. If the monkey broke fixation at any time during the trial, released the lever, or made an incor- rect lever turn, the trial was aborted immediately. The reward was given immediately after a correct lever turn however, the fixation spot and search array remained on for an additional 250���500 ms, and during this time the monkeys were free to make saccades without penalty. This was done to probe whether there were latent saccade plans that were being suppressed until after the reward. The intertrial interval from the re- moval of the visual stimuli and the reappearance of the fixation spot at the beginning of the next trial was at least 500 ms. Longer intertrial intervals occurred when the monkeys did not maintain gaze at the central location between trials or when the lever was not held in the vertical position. Monkey S was trained to report the location of the color singleton target of the search array. The stimuli were isoluminant green and red disks. The target could be either green or red, but within a block of trials, the color of the target and distractors did not change. Six stimulus loca- tions were used three were to the left and three were to the right of the fixation cross. A correct response was a lever turn corresponding to the location of the target stimulus relative to the fixation spot. Monkey C was trained to report the identity of a Landolt C among O distractors. The stimuli were gray rings, with one of them having a 0.5�� gap randomly on the left or right. Eight stimulus locations were used. A correct response was a lever turn corresponding to the location of the gap in the Landolt C. Data analysis. Lever position and eye position were sampled at 1000 Hz. Saccades were detected by using a computer algorithm that searched for elevated eye velocity ( 20��/s). Saccade initiations and terminations then were defined as the beginnings and ends of the monotonic changes in eye position that lasted at least 10 ms. A lever turn was defined as a turn 15�� from vertical. The beginning and end of each lever turn were de- fined as the beginning and end of the monotonic change in lever position before and after the 15�� threshold was reached. The time of the beginning of the lever turn on each trial was used as the reaction time for that trial. Activity recorded in the memory-guided saccade task was used for neuron classification. Activity was measured as a spike count per trial occurring in 150 ms time intervals. The visual response was measured between 50 and 200 ms after the target flash. Baseline activity was mea- sured during the last 150 ms before target presentation. The movement response was measured between 100 ms before and 50 ms after saccade initiation. Delay period activity was measured in a 150 ms interval of the delay period beginning 300 ms before the fixation spot disappeared, which cued the monkey to make a saccade to the remembered target location. The nonparametric Wilcoxon rank sum test was used to test for significant differences in spike counts across conditions. A neuron was defined as being visually responsive if the visual response was signifi- cantly greater than baseline activity ( p 0.05). A neuron was defined as being movement-related if the movement response was significantly greater than the late delay period activity. Neurons were classified as visual, visuomovement, or movement (see Fig. 5) based on these two statistical tests. A neuron was defined as being selective in the covert visual search task if the number of spikes per trial occurring during the interval from 100 to 250 ms after the presentation of the search array was significantly greater ( p 0.05) on trials in which the target of the search array fell in the receptive field of the neuron than on trials in which only distractors fell in the receptive field. The average spike density functions shown in Figures 3, 4, and 6 were obtained with a kernel that projects activity forward in time and approximates an EPSP (Thompson et al., 1996) and are used for viewing average spike activity only. Results In two monkeys, we controlled the locus of exogenously driven covert attention with a pop-out search task without eye move- ments (Fig. 1b). A salient oddball stimulus was presented among homogeneous distractors. The location of the target was random- ized from trial to trial. In this situation, the target stimulus popped out, automatically attracting attention (Theeuwes, 1994 Joseph and Optican, 1996 Egeth and Yantis, 1997 Nothdurft, 1999 Turatto and Galfano, 2000 Turatto et al., 2004). In addi- tion, the salient target stimulus was behaviorally relevant, which also encouraged subjects to focus attention to the location of the target for visual analysis (Nothdurft, 2002). The monkeys were trained to report with a manual lever turn either the location (monkey S) or the orientation (monkey C) of a singleton target among distractors without shifting gaze from a central fixation 9480 ��� J. Neurosci., October 12, 2005 ��� 25(41):9479 ���9487 Thompson et al. ��� Covert Attention in the Frontal Eye Field