Figure 2.

Patient Study: Trial Examples and Behavioral Data

(A) Memory-guided search task in experiments 1, 2, and 3. Participants were given either the name of a color cue, which always matched the search target (100% valid), or a neutral cue (“no cue”). This was followed by a search task that included finding the circle containing two gaps in the vertical plane and reporting whether it was located to the left or to the right of the central fixation. Participants responded via a button press.

(B) Delayed recognition task. In experiments 1 and 2, the ability to maintain the memory cue was assessed in a separate recognition task. Participants were required to remember the color word, and, following a 2 s delay, they were required to respond whether the colored circle matched or did not match the verbal cue. Experiments 3 and 4 incorporated a memory test after the search to ensure that the cue was held in memory across the delay.

(C) Invalid cueing in experiment 4. If the color word matched an item presented in the search display, this would never be the item surrounding the target (100% invalid trials).

(D) Experiment 1: Cue-validity effects on search (median neutral RT − median valid RT) for the thalamic patients and for the control groups (error bars show SEM of the cue-validity effect). The size of the cue-validity effects was reduced in the thalamic group relative to age-matched patients with lesions outside the thalamus (n = 18; t(22) = −3.5; p = 0.002; independent sample two-tailed t test) and in the nonstroke group of age-matched controls (n = 22; t(26) = −3.4; p = 0.002). Individual median RT data and SEM across the different validity conditions are presented in Table S2.

(E) Experiment 2: Search RTs as a function of cue-validity (Neu, neutral; Val, valid) and cue-search delay in patients VL1 and VL2. Note that performance of the patients in the short delay condition was compared with the performance of control groups from experiment 1, which had been tested with the same delay. Cueing effects differed significantly between the patients with lesions outside the thalamus and the two VL patients, and the same held for the comparison with the nonstroke group (t > 11, p < 0.0001).

(F) Experiment 3: Search RTs as a function of cue validity in the version of the task that incorporated a memory test after the search. VL patients showed reversed validity effects relative to the nonstroke controls and the stroke control group (all one sample t > 10, p < 0.0001).

(G) Experiment 4: Cue-invalidity effects (invalid RT − neutral RT) for the thalamic patients and the healthy controls. Importantly, we only analyzed search trials with correct recognition memory responses. Healthy controls (n = 11) displayed slower search in invalid trials relative to neutral trials (F(2,10) = 50.47; p = 0.0001). Invalidity effects were significantly lower in the thalamic patients than in the healthy controls (t(26) = −3.38; p = 0.002). This pattern of results held when the data were transformed to account for overall individual search latencies (Supplemental Experimental Procedures).