Data analysis

All eye movements with a minimum velocity of 200 °/s and a minimum duration of 30 ms were considered as saccades. The point in time when eye velocity passed the minimum velocity criterion determined saccade onset. Saccade offset was defined as the point in time

Distractor G/R ratio

0.18 0.24 0.37 0.62 1

Hit rate

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

contraversive target ipsiversive target all target positions

Fig. 2.4 Psychometric curves for distractor color determination in target-distractor trials.

Blue curve: trials with a contraversive target, red curve: trials with an ipsiversive target, black curve: all target positions combined, hit rate: proportion of trials with correct target identification among all trials with a valid response selection. The two distractor colors represented by distractor G/R ratios of 0.18 (dark orange, difficult distractor) and 1 (yellow, easy distractor) were later used in stimulation sessions.

2.2 Methods 21 when eye velocity dropped below 50 °/s. A trial was completed as soon as the animal either selected one of the peripheral stimuli or the fixation spot (see section 2.2.6). Hit rates were calculated as the proportion of correct trials among all completed trials. Reaction times (RTs) were measured as the time between target onset and saccade onset. Data analysis and statistical tests were performed using MATLAB R2014a (The MathWorks, Inc., USA). Trials were separated by the hemifield in which target or distractor were presented (contraversive or ipsiversive side relative to the side of stimulation) and averaged across all three contraversive and ipsiversive peripheral stimulus positions, respectively. Generally, behavioral effects were first tested using 2-way and 3-way repeated-measures ANOVAs (rmANOVAs, see below). Significant interaction effects were further investigated using paired-sample post-hoc t tests applying the False Discovery Rate (FDR) method to correct for multiple comparisons (Benjamini & Hochberg, 1995).

Response selection

To analyze how electrical microstimulation affected response selection (saccade to contraver-sive stimulus, saccade to ipcontraver-sivercontraver-sive stimulus, or central fixation) we calculated the probability of selecting a response among all completed trials for each experimental condition: single contraversive target, single ipsiversive target, single easy contraversive distractor, single difficult contraversive distractor, single easy ipsiversive distractor, single difficult ipsiversive distractor, contraversive target/easy ipsiversive distractor, contraversive target/difficult ip-siversive distractor, ipip-siversive target/easy contraversive distractor, ipip-siversive target/difficult contraversive distractor, double easy distractor, double difficult distractor, and double target, each condition without stimulation (control), with stimulation before the go signal, with stimulation at the go signal, or with stimulation after the go signal, respectively. Probabilities were calculated for each session separately and means and standard errors of means were calculated across sessions (n= 16).

Stimulation effects on response selection were investigated using three different rmANOVA designs. (1) Response selection in double-distractor and double-target trials were analyzed using a two-way rmANOVA with factors stimulus condition (easy double distractor, difficult double distractor, double target) and stimulation (control, before go, at go, after go).

(2) Response selection in target-distractor trials were analyzed using a three-way rmANOVA with factors stimulus condition (contraversive target/ipsiversive distractor, ipsiversive tar-get/contraversive distractor), perceptual difficulty (easy, difficult), and stimulation (control, before go, at go, after go). (3) Response selection in single-distractor and single-target trials were analyzed using a three-way rmANOVA with factors stimulus condition (single target, single easy distractor, single difficult distractor), hemifield of stimulus presentation

(contraversive, ipsiversive), and stimulation (control, before go, at go, after go). For con-ditions with two peripheral potential saccade options rmANOVAs were applied for each of the three possible response types separately (saccade to contraversive stimulus, saccade to ipsiversive stimulus, or central fixation). The rmANOVA for conditions with only one peripheral potential saccade option was applied on the probability of a saccade response.

Note that in these conditions the two possible response types are complementary to each other.

Thus, the rmANOVA on the probability of selecting the fixation response reveals the same results. Since our main interest was to investigate the effects of pulvinar stimulation starting at different time points relative to the go signal compared to the control condition without stimulation, significant interaction effects were further analyzed using paired-sample t tests for the three relevant comparisons between stimulation conditions (control vs. stimulation before go, control vs. stimulation at go, and control vs. stimulation after go) for each stimulus condition separately. Resultingpvalues were adjusted by FDR correction.

Saccade reaction times

To investigate whether electrical microstimulation delayed or facilitated saccade initiation we analyzed RTs in correct trials with saccades to targets for all experimental conditions except for conditions where no target was presented (single-distractor and double-distractor trials).

First, for each session RTs were averaged across all trials of the same experimental condition.

Then, for statistical analysis mean RTs and standard errors of means were calculated across sessions. In addition, in order to correct for general differences in the execution of saccades to contraversive and ipsiversive stimuli and to further quantify the magnitude of pulvinar stimulation effects on saccade RTs we subtracted control RTs from stimulation RTs for each stimulation condition per session and calculated mean RT differences and standard errors of mean RT differences across sessions for further statistical analysis.

Stimulation effects on mean RTs and mean RT differences were investigated using two different rmANOVA designs, respectively. (1) RTs in single-target trials were analyzed using a two-way rmANOVA with factors hemifield of stimulus presentation (contraversive, ipsiversive) and stimulation (control, before go, at go, after go). (2) RTs in double-target and target-distractor trials were analyzed using a three-way rmANOVA with factors stimulus condition (double target, target-distractor), hemifield of selected target (contraversive sac-cade, ipsiversive saccade), and stimulation (control, before go, at go, after go). The latter rmANOVA was applied for both levels of perceptual difficulty separately. Similar to the analysis of response selection, interaction effects were further analyzed using paired-sample t tests with FDR correction for the three comparisons between stimulation conditions.

2.3 Results 23