Experimental Protocol and Conditions

Experiments began with the isolation of single units based on visual responses to manually controlled light bars or RDPs of adjustable size, orientation and direction of motion while the animal fixated on a white fixation square and performed a change

detection task of a single moving RDP placed in the visual periphery. Once a visual response was obtained borders of isolated units were mapped manually and fitted with ellipsoids to obtain estimates of RF center position and size. These parameters were used to adjust the position and size of RDPs used to quantify directional and velocity tuning. For this purpose RDPs at twelve different directions and at least three different speeds were presented for more than one second each. Monkeys had to fixate and indicate a detection of a change in the direction of the tuning RDP to obtain juice/tea reward by releasing a lever. Responses to spiral-space motion were recorded while the animal detected a colour change of the fixation square.

We proceeded with the experiment only for well isolated units with a clear Gaus-sian shaped direction tuning curve. Preferred direction of motion, as well as the initial estimates of RF size and center position were used to adjust the experimental stimulus arrangement as described in the following sections.

2.2.1 Attentional RF Mapping

The spatial cueing task deployed in the current study is a variant of the attentional task used previously by Treue and Martinez-Trujillo (1999) (see figure 2.3, A): After obtaining fixation, the monkey had to touch a lever attached to the monkey chair and to encode the location of a stationary RDP presented for 440msec and serving as a spatial cue. After a brief delay (145msec) one target RDP at the cued location as well as two distracters were presented. After a further 160msec RF probe stimuli were sequentially onset at random locations at the intersections of a virtual grid of 43 to 54 positions covering the RF and its immediate surround. Probe duration was typically 187msec with an inter-probe blank interval of 27msec.

At random times during a trial targets and distracters changed direction of mo-tion once for 200ms. The monkey had to detect the direcmo-tion change of the target RDP while ignoring changes in the distracters. Direction changes could happen from 1200 to 5400msec during the course of the trial. The extent of the direction change was adjusted to the monkeys performance (depending on size and eccentricity of the target) and ranged from 30deg to 50deg. The monkey had to release a lever in response to the direction change of the target within a time window of 800ms.

Responses within 200msec after the target change were considered as anticipatory and counted as misses as were lever releases in response to direction changes of the distracters. Targets and distracter stimuli were always adjusted to move in the null direction of the isolated unit as determined by the direction tuning so as to evoke as little a response as possible in order to provide a modulatory range for responses to the RF probe stimulus. To reduce the influence of the attentional target and dis-tracter stimuli further, they were reduced in contrast. Because our primary interest

cue

Figure 2.3: Experimental design. A: Each trials began with a stationary RDP cueing the location of the later target stimulus position. When the lever was pressed in response to the cue and a brief delay was passed, the target RDP and two distracters were presented and began to move in the same direction. After a further delay probe RDPs of equal size and moving in the opposite direction of the continuously presented RDPs were sequentially flashed at the intersections of a virtual ellipsoid grid (indicated as grey dots) for 187 ms each, with an interprobe interval of 27 ms.

Tea reward was provided if the monkey indicated the detection of a brief change in the motion direction of the RDP at the target location while ignoring direction changes in the distracting RDPs. B: Time axis of stimulus on and offsets during a trial of the cue (bottom line), target and distractor stimuli (middle line) and RF probes (upper line). C: Typical arrangement and terminology of stimuli for the spatial attention task. On each trial one of the three stimuli were spatially cued to become the target while the remaining two stimuli remained as distracters. S1 and S2 were always placed within the RF, while S3 were placed in the opposite hemifield at a similar eccentricity as the RF stimuli. Grey dots indicate the position at which the RF probe stimulus appeared sequentially in random order.

was in the probe responses, the RF probes were always full contrast and adjusted to move in the neurons preferred direction. In order to obtain a baseline response for the RF maps, we always left one time slot of the probe sequence blank.

Spatial arrangement for RF mapping

Position and size of the target and distracters and the grid for the probes were adjusted with a custom program written with matlab (The MathWorks). The main rationale for the adjustment was to ensure (i) that two of the attentional stimuli lie within activating regions of the neuronal RF (ii) that both stimuli in the RF were positioned at equal eccentricity and (iii) that RF probes covered the RF plus its immediate surround. Therefore, the grid was centered at the estimated RF position and extended 1.5 to 2.5 times the RF radius of the longer axis of an ellipsoid RF estimate. In most experiments the grid had an ellipsoid shape with the longer axis running tangentially to the fixation point. Probe radius was always set to half the distance of adjacent grid positions. Target and distracters were always the same size

as the probes and were assigned to positions of the RF grid with an equal distance to the grid center and always located on the axis perpendicular to the fixation square.

This ensured identical eccentricity of the potential attentional targets within the RF. The third stimulus was positioned at a similar eccentric position in the opposite hemifield (cf. figure 2.3 C, as an example).

Attentional conditions

During the course of the experiment, conditions were randomly intermixed. In the terminology introduced in figure 2.3 B, the monkey was cued either to attend to S1, S2, or S3. S1 and S2 are located within the RF and conditions with attention allocated to them will thus be called the inside conditions, while attention to S3 represent the attend-outside condition. In addition to the three conditions with peripheral attention there was always a neutral condition indicated to the monkey by a red fixation square and the lack of the stationary spatial cue at the beginning of the trial. It was otherwise identical in visual stimulation during probe presentation. In this fixation with S1S2, i.e. with S1 and S2 within the RF, the monkey had to detect a luminance change of the red fixation square and thus did not require attention to a peripheral stimulus. The control condition was presented randomly interleaved with the peripheral attention conditions during the course of the experimental session.

Sensory RF mapping

Sensory RF profiles were also obtained in a further condition in which only probe stimuli were presented during the trial. For this sensory condition, (fixation probe condition in the following), the monkey had to detect a luminance change of the fixation square while the receptive field was probed with stimuli at all positions of the grid including those which were assigned to S1 and S2 in the conditions described above. Probes were in all respects identical to those described above. This sensory RF mapping is similar to mapping regimes reported in the literature and thus allows a comparison with previous studies.

2.2.2 Direction Tuning: Motion Reverse Correlation

In addition to the mapping conditions described above we also interleaved trials to monitor and validate the direction tuning of the isolated unit during attentional RF mapping. Direction tuning was based on the motion reverse correlation (MRC) method which is a time efficient way to obtain directional tuning and sensory re-sponse latencies of motion selective cells in motion selective cells in macaque area MT and cat striate cortex (Perge et al., 2002; Borghuis et al., 2003). We adapted the MRC method by creating RDPs which changed to one of twelve directions of motion every 80msec (6 frames), while leaving speed constant. The MRC stimulus was presented in trials requiring the monkey to fixate a red fixation square which

changed luminance at a random time. It was always positioned in the center of the estimated RF (the central grid position) between S1 and S2 with a radius 1.5 times that of S1, S2, and the probes.