Context Cue Dependent Saccadic Adaptation in Rhesus Macaques Cannot Be 1 Elicited Using Color

47 48 When the head does not move, rapid movements of the eyes called saccades are used 49 to redirect the line of sight. Saccades are defined by a series of metrical and kinematic 50 (evolution of a movement as a function of time) relationships. For example, the 51 amplitude of a saccade made from one visual target to another is roughly 90% of the 52 distance between the initial fixation point (T0) and the peripheral target (T1). However, 53 this stereotypical relationship between saccade amplitude and initial retinal error (T1-T0) 54 may be altered, either increased or decreased, by surreptitiously displacing a visual 55 target during an ongoing saccade. This form of motor learning (called saccadic 56 adaptation) has been described in both humans and monkeys. Recent experiments in 57 humans and monkeys have suggested that internal (proprioceptive) and external (target 58 shape, color, and/or motion) cues may be used to produce context dependent 59 adaptation. We tested the hypothesis that an external contextual cue (target color) could 60 be used to evoke differential gain (actual saccade/initial retinal error) states in rhesus 61 monkeys. We did not observe differential gain states correlated with target color 62 regardless of whether targets were displaced along the same vector as the primary 63 saccade or perpendicular to it. Furthermore, this observation held true regardless of 64 whether adaptation trials using various colors and intrasaccade target displacements 65 were randomly intermixed or presented in short or long blocks of trials. These results 66 are consistent with hypotheses that state that color cannot be used as a contextual cue 67 and are interpreted in light of previous studies of saccadic adaptation in both humans 68 and monkeys. 69 70 Introduction 71 72 Saccades have been frequently studied in an attempt to uncover the neural 73 mechanisms underlying the maintenance of movement accuracy and precision. 74 Furthermore, saccades have been frequently used to study the ability of primates to pair 75 arbitrary sensory stimuli (e.g. a visual object’s color, shape, orientation, and/or motion 76 properties) with a specific motoric response (for reviews see: Hopp & Fuchs, 2004; 77 Iwamoto & Kaku, 2010; Pélisson, Alahyane, Panouillères, Tilikete, 2010; Herman, 78 Blangero, Madelain, Khan, Harwood, 2013; Liversedge, Gilchrist, & Everling, 2011; 79 Gold & Shadlen, 2007; Shadlen & Kiani, 2013). Saccades are rapid, conjugate eye 80 movements that may be used to reorient the line of sight such that the high-resolution 81 region of the retina (the fovea) can be aligned with objects of interest (Leigh & Zee, 82 1999). Primate saccades are defined by a series of stereotypical metrical relationships 83 between amplitude, peak velocity, and duration (Bahill, Clark, & Stark, 1975; Baloh, 84 Sills, Kumley & Honrubia, 1975; vanGisbergen, Van Opstal, & Ottes, 1984) and primary 85 saccade gain (movement amplitude/initial retinal error) is approximately 0.90-0.95 86 (Hyde, 1959; Becker & Fuchs, 1969; Becker, 1972; Henson, 1978, 1979; Prablanc, 87 Masse, and Echallier, 1978; Kowler & Blaser, 1995). Modifications to these relationships 88 may result from neuromuscular disease (Leigh & Zee, 1999) or, in the case of saccade 89 gain, by experimental manipulations. 90 Under laboratory conditions, motor learning in the saccadic system (“saccadic 91 adaptation”) has mostly been studied by surreptitiously introducing a visual error at the 92 end of a saccade that was aimed at a target located at a particular vector (magnitude 93 and direction) relative to where the subject was fixating (Albano, 1996; McLaughlin, 94 1967; Deubel, 1987, 1991; Deubel, Wolf, & Hauske,1986; Miller, Anstis, & Templeton, 95 1981; Noto, Watanabe, Fuchs,1999; Robinson, Noto, & Bevans, 2003; Scudder, 96 Batournia, & Tunder, 1998; Semmelow, Gauthier, & Vercher, 1987; Straube, Fuchs, 97 Usher, Robinson, 1997). Under these circumstances: 1) changes in primary saccade 98 amplitude follow a roughly exponential time course with “rate constants” around 30–60 99 saccades in humans (Albano, 1996; Deubel et al, 1986; Deubel, 1987) and 100-800 100 saccades in monkeys (Straube et al, 1997); 2) the change in primary saccade gain is 101 appropriate to reduce the visual error induced at the end of the movement, but is rarely 102 large enough to consistently place the fovea on the location of the target after it was 103 displaced during the saccade; 3) the magnitude of backward adaptation is larger than 104 forward adaptation in response to the same post-saccadic visual error (Straube et al, 105 1997); 4) adaptation effects transfer to saccades with vectors similar to those initially 106 adapted (“adaptation fields”; Noto et al, 1999); and 5) a corrective movement is not 107 necessary for learning to progress (Wallman & Fuchs, 1998), which indicates that a 108 visual error signal is sufficient enough to drive this form of motor learning. 109 There has recently been a great deal of interest in “context dependent” saccadic 110 adaptation (see Pélisson et al, 2010, Herman et al, 2013 for reviews). In this type of 111 experiment a cue, either internal (e.g. proprioceptive feedback of eye position) or 112 external (e.g. visual target properties), is used during and after the learning process to 113 elicit different gain states. For example, Alahyane & Pélisson (2004) have shown that 114 human horizontal saccade amplitude can be simultaneously reduced and increased to 115 the same initial target displacement (T1-T0) depending upon the vertical orbital eye 116 position (either 12.5° up or 25° down) at which a subject initiates the primary saccade. 117 On the other hand, differential changes in saccade gain based on eye position have 118 proven difficult to elicit in rhesus monkeys under similar conditions (Figures 2 & 8 in 119 Tian & Zee, 2010). 120 The use of external visual cues to drive different gain states during saccadic 121 adaptation has led to a mixture of results in human subjects. Deubel (1995), in a few 122 human subjects, was unable to elicit distinct gain states during horizontal adaptation 123 using colored, static targets (red crosses and green circles) that were displaced by a 124 few degrees of visual angle along the same axis as the primary saccade. Subsequent 125 investigations of human saccadic adaptation were also unable to elicit context 126 dependent adaptation using either static shapes (diamonds versus squares; Bahcall & 127 Kowler, 2000) or moving visual targets with different shapes and colors (Azadi & 128 Harwood, 2014). However, other explorations of saccadic adaptation have shown that 129 visual cues such as flickering versus non-flickering targets (Herman et al., 2009), yellow 130 squares versus green circles (Madelain et al., 2010), and other properties of moving 131 targets (speed and direction) (Azadi & Harwood, 2014) can be used to elicit different 132 gain states. 133 Given the different behaviors of human subjects in context cue adaptation 134 experiments using different visual stimuli, and the potential for monkeys to have 135 different behavior to the same stimuli that elicited context dependent adaptation 136 behavior in humans, we designed the current study to test the hypothesis that color can 137 be used as a contextual cue in a saccade adaptation task. Although our results concur 138 with those observations made by Deubel (1995) and Azadi & Harwood (2014), our 139 experiments provide a more extensive data set to support the assertion that the 140 saccadic adaptive control system does not differentiate between color targets of the 141 same shape. These data are interpreted in the light of previous context dependent 142 adaptation studies. 143 144 Materials and Methods 145 146 All procedures were approved by the Institutional Animal Care and Use 147 Committee of the University of Pittsburgh and were in compliance with the guidelines 148 set forth in the United States Public Health Service Guide for the Care and Use of 149 Laboratory Animals. Three male rhesus monkeys (BB, BU, & WE) weighing 7.0 -13.0 kg 150 served as subjects. Each monkey had a small head-restraint device secured to the skull 151 during an aseptic surgery. In an additional aseptic surgery, monkeys BB and BU had a 152 scleral coil implanted for monitoring gaze position (Judge et al. 1980). After full 153 recovery, each animal was trained to sit in a primate chair with their head restrained and 154 a sipper tube was placed near the mouth for reward delivery. Subjects were 155 subsequently trained to make gaze shifts to visual targets, but were not used for 156 adaptation studies prior to their participation in the current series of experiments. 157 For monkeys BB and BU, visual stimuli, behavioral control, and data acquisition 158 were controlled by a custom-built program that uses LabVIEW architecture on a real159 time operating system supported by National Instruments (Austin, TX) (Bryant and 160 Gandhi 2005). These animals sat inside a frame containing two alternating magnetic 161 fields that induced voltages in the eye coil and thus permitted measurement of 162 horizontal and vertical eye positions (Robinson 1963). Visual targets were displayed on 163 a 55 inch, 120 Hz resolution LED monitor. For monkey WE, eye movements were 164 monitored using an eye tracker (Eye Link 1000, SR Research Ltd, Mississauga, 165 Ontario, Canada) and visual targets were presented on a 21 inch, 100 Hz resolution 166 CRT monitor. 167 168 General Chronology of Experimental Sessions, Trial Types, and Reward Criterion 169 170 Every adaptation session had at least three phases that occurred in the following 171 chronological order: 1) a “pre-adaptation phase” in which only probe trials were 172 presented; 2) an “adaptation phase in which only adaptation trials were presented; and 173 3) a “post-adaptation phase” in which only probe trials were presented. Both probe and 174 adaptation trials (Figure 1A, B) began with the illumination of an initial fixation target 175 (T0). Subjects were required to look at and maintain fixation of T0 for a variable duration 176 (500-1000ms, 50 or 100ms increments). Trials were aborted if the line of sight deviated 177 beyond a computer-defined window (3° radius) surrounding T0. However, if fixation was 178 maintained, a peripheral target (T1) was illuminated. T0 and T1 overlapped between 0 to 179 750ms in 250ms increments. If the subject continued to maintain fixation of T0 for the 180 duration of this overlap period, then T0 was extinguished, cuing the animal to make a 181 saccade to T1. To this point both probe and adaptation trial types were identical. During 182 “probe” trials (Figure 1A), T1 remained illuminated until the position of the line of sight 183 exited the computer window centered on the location of the no longer visible T0. T1 was 184 turned off after the line of sight crossed this position criterion during probe trials; 185 therefore, no visual feedback was available on these trials. “Adaptation” trials (Figure 186 1B) were similar to probe trials except that after T1 was extinguished, a target (T2) was 187 immediately illuminated in a new spatial location. The new location could be further 188 away from (forward adaptation) or closer to (backward adaptation) T0 along the 189 horizontal meridian during trials attempting to elicit adaptation of the horizontal 190 component of the primary saccade vector (Figure 1C). In experiments attempting to 191 elicit changes in the vertical component of the primary saccade vector, targets could be 192 displaced above (upward adaptation) or below (downward adaptation) the horizontal 193 meridian (Figure 1D). The latter experiments are hereafter referred to as “orthogonal 194 adaptation” experiments. Lastly, note that the horizontal direction (left or right) of the 195 primary saccade during adaptation trials was pseudorandomly alternated between 196 sessions. 197 During the preadaptation phase, the locations of T1 always included the T1 used 198 during adaptation trials and typically the T2 locations and a target location that would 199 evoke a primary saccade in the direction opposite that produced during adaptation trials 200 (for specific examples of targets, see Hypotheses, Predictions, and Experimental 201 Session Subtypes below). Subjects were rewarded for entering a window surrounding 202 T1 that had a radius of 6o and remaining in the window for a minimum of 250ms. During 203 the adaptation phase, the reward window associated with T1 was elliptical, centered on 204 a location between T1 and T2, and enlarged such that it encompassed and extended 205 beyond these targets by ~5o. The reward window associated with the T1 used during 206 adaptation trials remained enlarged during the post-adaptation probe phase and 207 subjects needed to maintain fixation within this window for a minimum of 250ms in order 208 to be rewarded. The reward criterion for other T1 locations during the post-adaptation 209 phase was the same as that used during the pre-adaptation phase. 210 In each of our color cue experiments, the targets (T0, T1, T2) were red, green, or 211 yellow dots subtending ~1° of visual angle. Target color remained constant within a 212 given trial and trials using a particular color were randomly intermixed throughout all 213 phases of the experiment. Therefore, during adaptation trials, the fixation point color 214 could indicate forward (green) or backward (red) adaptation. Target locations during 215 probe and adaptation trials remained fixed within a data collection session. 216 The exact number of probe trials during the preand post-adaptation epochs 217 varied depending upon the number of colors used in a given experiment (one versus 218 three), the type of experiment and thereby the number of potential T1 and T2 locations 219 used during adaptation trials, and the willingness of the subject to sustain effort for long 220 periods during the post-adaptation phase. Furthermore, we attempted to keep the 221 length of the adaptation phase relatively constant across experimental sessions such 222 that the subject experienced 450 or more adaptation trials per condition (900+ 223 adaptation trials total per session). Hence, the duration of the adaptation phase was 224 relatively constant across sessions within a subject. 225 226 227 228 Data Acquisition and Analysis 229 230 For both animals, each trial was digitized and stored on the computer’s hard disk for off231 line analysis in MATLAB (R2013b; Natick, Massachusetts, U.S.A). Horizontal and 232 vertical eye positions were stored with a resolution of 1ms. Component velocities were 233 obtained by differentiating the eye position signals. The onset and offset of saccades 234 were identified using a velocity criterion of 40°/second. Saccade amplitude was defined 235 as the change in eye position from the beginning to end of the eye movement based on 236 this velocity criterion. All of the data reported here are the result of measuring the first 237 (“primary”) saccade made towards T1 that was associated with the adaptation trials for a 238 particular session. This has traditionally been used to assess the subject’s motoric state 239 at various time points during adaptation experiments (see Hopp & Fuchs (2004) for 240 review). The primary saccade needed to occur within 500ms after the offset of T0 and 241 have a horizontal amplitude greater than 5° in order to be considered for further 242 analysis. 243 244 Saccade gain was defined by the following formula: 245 246 Saccade Gain = Saccade Amplitude 247 T1-Initial Eye Position 248 249 By definition a saccade with a gain that is less than 1.0 is hypometric whereas a 250 saccade with a gain greater than 1.0 is hypermetric. During color context cue 251 experiments, trials were then parsed based on whether the target color within a trial was 252 red, green, or yellow. For each color, the average gain or amplitude of saccades made 253 during all pre-adaptation phase probe trials using the T1 that was used during 254 adaptation trials (“pre-probe trials”) was compared with average of an equal number of 255 probe trials presented after a particular adaptation segment (“post-probe” trials). Across 256 all color cue experiments, the mean (±SD) number of preand post-adaptation trials 257 used for comparison was 22.9 ± 6.0 (range = 12:42) and was not different across colors 258 (green = 22.9±6.1; yellow = 23.2±6.2; red = 22.1± 5.6; p = .8106 Kruskal-Wallis test). 259 The difference between preand post-adaptation means for a given color within an 260 experimental session was compared using a Wilcoxon rank sum test. A Kruskal-Wallis 261 test was used to assess the effect of color on gain across multiple sessions. All tests 262 were done in the MATLAB Statistical Toolbox; significance was determined using a 263 Bonferroni corrected p-value (p < 0.0167) in the case where three means were 264