Reward prediction based on stimulus categorization in primate lateral prefrontal cortex Xiaochuan Pan, Kosuke Sawa, Ichiro Tsuda, Minoro Tsukada, Masamichi Sakagami Supplementary Information This PDF file includes Supplementary Figure 1 TO 8 Supplementary Table 1 and 2
Supplementary Figure 1 a Spikes s 1 c 6 4 2.6.4.2 R type n = 17 Non-preferred reward 5 5 1, Τime from first cue onset (ms) R type n = 34 Monkey Z 2 4 6 8 1 SPAT order in block b d.6.4.2.8.6.4.2 R type n = 46 1 2 3 4 5 SPAT order in block R type n = 27 Monkey H Monkey T Non-preferred reward 2 4 6 8 1 SPAT order in block Supplementary Figure 1 Population histagrom of R type neurons and the averaged activity in the first cue period as a function of SPAT order in block. (a) Population histogram of all R type neurons from the three monkeys, which was sorted by preferred reward condition (orange curve) and non-preferred reward condition (blue curve). (b) (d) the activity of R type neurons in the first cue was plotted as a function of SPAT order for the three monkey: (b) Monkey H, (c) Monkey Z and (d) Monkey T. Activity was sorted by two conditions: preferred reward condition (orange curve) and non-preferred reward condition (blue curve). The statistical significance was checked by Mann-Whitney test (two-tailed, p<.1). Error bars indicate s.e.m.
Supplementary Figure 2 a Large reward preferred group Spikes s 1 6 4 2 n = 49 Large rew, pref. stim Large rew, non. stim Small rew, pref. stim Small rew, non. stim 5 5 1, Time from first cue onset (ms) b 6 Small reward preferred group N=27 Spikes s 1 4 2 5 5 1, Time from first cue onset (ms)
Supplementary Figure 2 c Monkey H d Monkey Z Spikes s 1 6 4 2 n = 31 8 6 4 2 n = 25 5 5 1, Time from first cue onset (ms) 5 5 1, Time from first cue onset (ms) e Spikes s 1 6 4 2 n = 2 Monkey T Pref. rew, pref. stim Pref. rew, non. stim Non. rew, pref. stim Non. rew, non. stim 5 5 1, Time from first cue onset (ms) Supplementary Figure 2 Population histogram of neurons. (a) (b) The population histogram of neurons separated into two subgroups: one preferred the large reward (a) and the other preferred the small reward (b). All trials were sorted by four conditions: large reward with preferred stimulus (solid orange curve), large reward with non-preferred stimulus (dashed orange curve), small reward with preferred stimulus (solid blue curve) and small reward with non-preferred stimulus (dashed blue curve). (c) (e) The population histogram of cells for the three monkeys, respectively: (c) Monkey H, (d) Monkey Z and (e) Monkey T. Trials from each neuron were sorted by four conditions: preferred reward condition and preferred stimulus (solid orange curve), preferred reward condition and non-preferred stimulus (dashed orange curve), non-preferred reward condition and preferred stimulus (solid blue curve) and non-preferred reward condition and non-preferred stimulus (dashed blue curve). The black line marks the time of first cue onset. The gray area indicates the first cue period that was used for analysis of neuronal activities (Monkey H and Z, 1 4ms from the first cue onset, and Monkey T, 1 5ms from the first cue onset).
Supplementary Figure 3 a b c.6.4.2.6.4.2 n = 31 1 2 3 4 5 n = 31 Monkey H Monkey Z Monkey T Preferred stimulus Preferred stimulus Preferred stimulus Non-preferred reward 1 2 3 4 5.8.6.4.2.8.6.4.2 n = 25 2 4 6 8 1 d e f n = 25 2 4 6 8 1.8.6.4.2.8.6.4.2 n = 2 n = 2 2 4 6 8 1 Non-preferred stimulus Non-preferred stimulus Non-preferred stimulus Non-preferred reward 2 4 6 8 1 SPAT order in block SPAT order in block SPAT order in block Supplementary Figure 3 Averaged activity of neurons in the first cue period is plotted as a function of SPAT order in blocks. (a) (c) The activity based on the preferred stimulus for the three monkeys. (d) (f) The activity based on the non-preferred stimulus for the three monkeys, respectively. The orange curves indicate the data from the preferred reward condition, and the blue curves represent the data from the non-preferred reward condition. The statistical significance was checked by Mann-Whitney test (two-tailed, p<.5, p<.1). Error bars indicate s.e.m.
Supplementary Figure 4 a Monkey H b Monkey T Correct rate 1..9.8.7.6.5 ABC sequence Large reward Small reward 1..9.8.7.6.5 ABC sequence Large reward Small reward 2 4 6 8 1 BCA sequence BCA sequence Correct rate 1..9.8.7.6 1..9.8.7.6.5.5 2 4 6 8 1 1. CAB sequence 1. CAB sequence Correct rate.9.8.7.6.9.8.7.6.5 SPAT order in block.5 2 4 6 8 1 SPAT order in block Supplementary Figure 4 The correct rate of the first choice in SPATs with three sequences: ABC sequence (upper panel), BCA sequence (middle panel) and CAB sequence (bottom panel). (a) The data from Monkey H (left column) and (b) from Monkey T (right column). The orange curves indicate the correct rates in large reward trials, and the blue curves in small reward trials. Two monkeys showed similar reward modulated behavior with three sequences. After reward instruction trials, the correct rate in large reward condition is significantly higher than that in small reward condition from the first SPATs with all three sequences (Mann-Whitney test, two-tailed, p<.5, p<.1). Error bars: s.e.m.
Supplementary Figure 5 a Monkey H b Monkey T ABC sequence ABC sequence.6.4.2 R type n = 21 Non-preferred reward.6.4.2 R type n = 24 Non-preferred reward 2 4 6 8 1 BCA sequence BCA sequence.6.4.2.6.4.2 2 4 6 8 1.6.4.2 CAB sequence SPAT order in block.6.4.2 CAB sequence 2 4 6 8 1 SPAT order in block Supplementary Figure 5 Averaged activity of R type neurons in the first cue period is plotted as a function of SPAT order in blocks with three sequences. (a) The data from Monkey H with three sequences (upper panel: ABC sequence, middle panel: BCA sequence and bottom panel: CAB sequence). (b) The data from Monkey T with three sequences (upper panel: ABC sequence, middle panel: BCA sequence and bottom panel: CAB sequence). Activity was sorted by two reward conditions: preferred reward (orange curves) and non-preferred reward (blue curves). The statistical significance was checked by Mann-Whitney test (two-tailed, p<.5, p<.1). Error bars: s.e.m.
Supplementary Figure 6 Spikes s 1 8 6 4 2 a Monkey H ABC sequence n = 14 Pref. rew, pref. stim Pref. rew, non. stim Non. rew, pref. stim Non. rew, non. stim 6 4 2 b Monkey T ABC sequence n = 16 5 5 1, 5 1, 8 BCA sequence BCA sequence Spikes s 1 6 4 2 4 2 5 5 1, 5 1, 8 CAB sequence 4 CAB sequence Spikes s 1 6 4 2 2 5 5 1, Time from the cue onset (ms) 5 1,
Supplementary Figure 6 c Category index between ABC and CAB sequences 1.8.6.4.2.2.4.4.2.2.4.6.8 1 Category index between ABC and BCA sequences Supplementary Figure 6 Population histograms of neurons with three sequences and scatter diagram of category index. (a) Histogram of Monkey H with three sequences: ABC (upper panel), BCA (middle panel) and CAB (bottom panel) sequences. (b) Histograms of Monkey T with three sequences: ABC (upper panel), BCA (middle panel) and CAB (bottom panel) sequences. Figure formats and labels are the same as in Supplementary Fig. 3. (c) The scatter diagram of category index between ABC and BCA sequence versus category index between ABC and CAB sequences.
Supplementary Figure 7 Monkey H a ABC sequence BCA sequence CAB sequence b.8.6.4.2.8.6.4.2 Preferred stimulus Preferred stimulus Preferred stimulus n = 14 Non-preferred stimulus n = 14 Non-preferred reward Non-preferred reward.8.6.4.2.8.6.4.2 Non-preferred stimulus.6.4.2.6.4.2 Non-preferred stimulus SPAT order in block SPAT order in block SPAT order in block
Supplementary Figure 7 c d.8.6.4.2.8.6.4.2 ABC sequence Preferred stimulus 2 4 6 8 1 2 4 6 8 1 BCA sequence Preferred stimulus CAB sequence Preferred stimulus Non-preferred stimulus Non-preferred stimulus Non-preferred stimulus n = 16 n = 16 Non-preferred reward.8.6.4.2.8.6.4.2 Monkey T 2 4 6 8 1 SPAT order in block SPAT order in block SPAT order in block 2 4 6 8 1.8.6.4.2.8.6.4.2 2 4 6 8 1 2 4 6 8 1 Supplementary Figure 7 Averaged activity of neurons in the first cue period is plotted as a function of SPAT order with three sequences. (a). The data of Monkey H for the preferred stimulus with three sequences (ABC sequence: left column, BCA sequence: middle column and CAB sequence: right column). (b) The data of Monkey H for the non-preferred stimulus with three sequences (ABC sequence: left column, BCA sequence: middle column and CAB sequence: right column). (c) The data of Monkey T for the preferred stimulus with three sequences (ABC sequence: left column, BCA sequence: middle column and CAB sequence: right column). (d) The data of Monkey T for the non-preferred stimulus with the three sequences (ABC sequence: left column, BCA sequence: middle column and CAB sequence: right column). Activity was sorted by two reward conditions: preferred reward (orange curves) and non-preferred reward (blue curves). The statistical significance was checked by Mann-Whitney test (two-tailed, p<.5, p<.1). Error bars: s.e.m.
Supplementary Figure 8 a The MRI image of the brain surface and outlines of the grid Center line of the chamber The schematic of the grid 1mm 1mm 1mm Lateral Middle X X 4 mm Principal sulcus Arcuate sulcus 3 mm 1mm 1mm 1mm
Supplementary Figure 8 b Right Hemiphere Monkey H Left Hemiphere Arcuate sulcus Principal sulcus c Monkey Z Right Hemiphere Arcuate sulcus Left Hemiphere Principal sulcus d Monkey T Arcuate sulcus Left Hemiphere Principal sulcus Recorded track One R type cell Two R type cells Three R type cells One cell Two cells Three cells 1mm
Supplementary Figure 8 Anatomical location of recording sites and the distribution of R type and neurons in LPFCs for the three monkeys. (a) The MRI image of the brain surface of Monkey H's left hemisphere with the chamber's position (left figure). The length of the chamber (posterior-anterior direction) was 4 mm, and the width (middle-lateral direction) was 3 mm. In order to get the brain surface image relative to the chamber position, a grid (the same size as the chamber, as indicated by right figure) was installed on the chamber during scanning image. The black thick lines in the grid indicate walls that were filled with magnevist liquid inside (enhance the MRI signal, corresponding to the black outlines in the image figure). The surface MRI image was oriented 25 degree; parallel to the chamber's orientation (middle-lateral direction). The red mark "x" in the image figure indicates the center position of the chamber. (b) (d) The distribution of recording sites and R and s in Monkey H (b), Monkey Z (c) and Monkey T (d). The empty black: recorded sites at which no reward-related neurons were found; the filled red circle and the empty blue triangle: recording sites at which R and neurons were found.
Supplementary Table 1 The correct rates in first SPATs and later trials First SPATs Later trials Large reward Small reward Large reward Small reward (mean ± sem) (mean ± sem) (mean ± sem) (mean ± sem) Monkey H 94.4% ±.7% 81.5% ± 1.6% 93.5% ±.7% 83.3% ± 1% Monkey Z 81.2% ± 2.2% 69.6% ± 2.3% 85.6% ± 1% 67.7% ± 1.3% Monkey T 92.4% ± 1.2% 77.9% ± 1.8% 84.8% ± 1% 69.8% ± 1% This table shows the performance in first sequential paired-association trials (SPATs) (the first trials just after reward instruction in blocks) and later trials (including all remaining trials in blocks) for three monkeys (H, Z and T). A two-way ANOVA (trial order (first: later) vs. reward (large: small)) revealed that Monkey H had a significant main effect of reward on performance (p<.1), there was no significant main effect of trial order (p>.2) and no interaction (p>.6); Monkey Z had a significant main effect of reward (p<.1), there was no significant main effect of trial order (p>.4) and no interaction (p>.8); Monkey T showed both a main effect of reward (p<.1) and a main effect of trial order (p<.1), but no interaction (p>.3).
Supplementary Table 2 The correct rates of first SPATs in switched-reward condition and constant-reward conditions Switched-reward condition Constant-reward condition Large reward Small reward Large reward Small reward (mean ± sem) (mean ± sem) (mean ± sem) (mean ± sem) Monkey H 94.1% ±.9% 83.2% ± 1.4% 95.2% ±.8% 76.4% ± 3.1% Monkey Z 84.3% ± 2.1% 68.7% ± 2.4% 8.3% ± 5.9% 62.2% ± 6.7% Monkey T 92.3% ± 1.3% 78.5% ± 2.2% 9.4% ± 2.8% 84.% ± 3.1% The table shows the performance of first sequential paired-association trials (SPATs) in switched-reward condition (the stimulus-reward contingency was reversed between two consecutive blocks) and constant-reward condition (the stimulus-reward contingency was remained between two consecutive blocks). The statistic significance was checked between large reward and small reward trials by two-tailed t-test (: p<.1; : p<.5).