Supplemental Figure 1. Examples of entrainment, attentional phase shifts and associated phase-locked activity between .7 and 60 Hz

Locations of the all examples are indicated in Supplemental Figure 5. All those electrodes showed a significant entrainment to at least one of the stimulus streams and a significant effect of attention on the phase of the entrained delta rhythm. First four examples were recorded from different parts of the frontal cortex (motor cortex excluded). Electrode LGd43 in patient L4 was implanted over the inferior/middle frontal gyrus, LFA2 and LFx4 in patient L5 over the anterior part of the orbito-frontal cortex and electrode PSF3 in patient S6 over the ventral/posterior orbito-frontal cortex. All those electrodes, like most inferior frontal electrodes showing a significant attentional phase shift (see Supplemental Figure 3) showed a phase difference between 'Attend Auditory' and 'Attend Visual approaching phase opposition. There was little or no phase-locked activity in frequencies above 1.5 Hz suggesting that entrainment was not due to evoked activity at the rate of stimulation. Electrodes LP3 in patient L4 and PG52 in patient S5 were implanted over between the frontal and the parietal cortex, close to the sensory/motor representation of the hand. Electrical stimulation at electrode PG52 elicited a deficit in a hand motor task (see Supplemental Figure 5) suggesting it might indeed cover the motor cortex. Electrode LP3 was not tested. Neither electrode shows strong phase-locked activity except at 1.5 Hz. The value of the phase shift was close to 180º for electrode LP3, whereas it was close to 45º for electrode PG52. Note that the only other electrode showing a significant attentional phase effect and positively tested in a hand motor task (see Supplemental Figure 5) showed an attentional phase shift larger than 90º. Electrodes LGp5 in patient L5 and G36/G44 in patient S4 were implanted over the posterior superior temporal cortex/ angular gyrus. Electrode G44 was positively tested for language (Supplemental Figure 5), electrode G36 was cleared and electrode LGp5 was not tested. Those 3 electrodes showed a close-to-180º attentional phase difference in the delta band and negligible phase-locked activity above 1.5 Hz, like most significant electrodes located in the same area (Supplemental Figure 4).

Supplemental Figure 2. Examples of entrainment, attentional phase shifts and associated phase-locked activity between .7 and 60 Hz

Locations of the all examples are indicated in Supplemental Figure 5. All those electrodes showed a significant entrainment to one of the stimulus streams and a significant effect of attention on the phase of the entrained delta rhythm. First three examples are likely to be over/in the auditory cortex. Electrodes RPLT56 in patient S2 and TG31 in patient S3 were probably implanted over the superior temporal cortex. Electrode RAD8 was a depth electrode in the lateral part of the superior/middle temporal lobe, possibly close to the STS. Electrode TG31 was cleared during the cortical stimulation mapping; the two other electrodes were not tested (Supplemental Figure 5). Like many electrodes in this area, electrodes RPLT56 and TG31 showed a small (but significant) attentional effect on the phase of the entrained delta (see also Supplemental Figure 3) and important phase-locked activity following auditory sensory stimulation (see also Supplemental Figure 4). The large phase-locked activity extending from 1.5 to 25 Hz suggests that at least part of the delta entrainment was due to a steady state evoked response at the stimulation rate. Electrode RAD8 did not show as strong a phase-locked activity but it was stronger following auditory than visual stimulation, suggesting that it also recorded from the auditory cortex although it was further from the primary auditory cortex than RPLT56 and TG31. The associated attentional phase shift was larger than 90º. Electrodes LGd30 in patient L5 was either over the superior temporal cortex or the facing mouth/tongue sensory/motor cortex. Electrical stimulation at this electrode elicited a sensation in the tongue which could suggest it was closer to the parietal cortex (Supplemental Figure 5). However, we also recorded a strong phase-locked response to auditory stimulation. Since our recording was monopolar, an electrode over the parietal cortex can record auditory activity from the superior cortex. Accordingly, the attentional phase effect was significant but less than 45º. The following 3 electrodes were implanted over the motor/frontal cortex. Electrodes FG7 in patient S5 and TG6 in patient S3 were positively tested in a mouth/tongue motor task (Supplemental Figure 5), which suggests they were implanted over the motor cortex. Electrode G41 in patient S4 was positively tested for language, which suggests that it was implanted close to Broca’s area. All three electrodes recorded somewhat larger phase-locked activity following auditory stimulation than following visual stimulation, suggesting they also recorded from the auditory cortex. Like other electrodes over this area (Supplemental Figure 3), the value of the attentional phase shift ranged from close to 0º to close to 180º. Electrode G23 in patient S4 was implanted over the posterior parietal cortex and was cleared during the cortical stimulation mapping. It showed a fairly large attentional delta phase shift and negligible auditory and visual phase-locked activity. Electrodes RTP2 in patient L5 and PSO6 in patient S6 were implanted over different parts of the visual cortex. Electrode RTP2 was probably implanted over a fairly early area of the visual cortex and PSO6 was located in the ventral pathway, sub-occipitally. PSO6 was cleared during the stimulation mapping and RTP2 was not tested. Both electrodes showed a strong phase locking to visual stimulation at 1.5 Hz and higher suggesting that part of the entrainment was due to sensory steady-state activity (see also Supplemental Fig. 3 and 4). The attentional phase shift was significant and close to 0º, like most electrodes in occipital regions (see Supplemental Figure 4).

Supplemental Figure 3. Activity phase-locked to auditory and visual stimulations in the auditory and visual attention conditions.

a. Each sphere corresponds to an electrode and its diameter and color codes for the proportion of significant phase locking values in a time-frequency window between 50 and 350ms and 3 and 30 Hz. Phase-locked activity to auditory stimuli (2 left panels) were found around the superior temporal cortex but could be consistently recorded from parietal and frontal areas across the sylvian fissure. By transparency, one can also see important phase-locked activity recorded from depth electrodes around the STS. Phase-locked activity to visual stimuli (2 right panels) was seen over the posterior dorsal and the ventral (fusiform gyrus) occipital cortices. There was some phase-locked activity over the motor cortex as well. Some areas showed an effect of attention: activity phase-locked to auditory stimulation seemed stronger in depth electrodes when attention was paid to auditory stimuli. Visual phase-locked activity activation extended more anteriorly towards the temporal cortex (both ventrally and laterally) when patients paid attention to visual stimuli. Effects of attention in other areas were more inconsistent, some electrodes showing increases and other electrodes showing decreases in phase-locked activity when attention was paid to the auditory or the visual stream. Locations of the electrodes was obtained by co-registration of the post-implant CT scan to the pre-implant structural MRI of each patient, Talairach normalization to the coordinate system of the MNI single subject MRI and manual adjustment to match functional data from the cortical mapping and additional anatomical information. The maximum proportion of significant PLVs, indicated by dark red sphere was 81 %.

b. Same as a, but using the hue/contrast color scaling defined in the manuscript to emphasize the effect of attention. Activity locked to auditory stimuli was more extended over the temporal cortex when patients were paying attention to auditory stimuli. Activity locking to visual stimuli was stronger over the fusiform gyrus and over the ventral motor cortex when patients were paying attention to visual stimuli. The hue of each density map was estimated by a linear combination of 3D Gaussian density function centered at each electrode location, with a width proportional to the proportion of significant PLVs in the 3-30Hz/50-350ms time-frequency window following stimulation. The value of the maximum hue and saturation values across the 4 conditions was normalized to 1. the width of the Gaussian density function at each electrode was multiplied by the proportion of significant PLVs (the phase-locking index), normalized and converted to the composite hue/brightness color map.

Supplemental Figure 4. Size of attentional phase shift and correlation with sensory phase-locked activity

The diameter and the color of the spheres are proportional to the maximum, between 'Auditory-locked' and 'Visual-locked' analysis, of the absolute difference of the entrained delta rhythm phases between the ‘Attend Auditory’ and the ‘Attend Visual’ conditions. a only displays electrode with a significant phase shift after correction, b displays electrodes with significant phase shift without correction and c all analyzed electrodes (with a significant entrainment of the 1.5 Hz rhythm). Small attentional phase shifts were consistently found over the ventral and posterior occipital cortex, the superior temporal cortex, the inferior somatosensory/motor cortex. Intermediate values were consistently found over the superior part of the frontal cortex. Large values approaching phase opposition were consistently found over the posterior superior temporal/angular gyrus and over the orbito-frontal cortex. Other parts of the frontal cortex and the parietal cortex showed both small and large attentional phase shifts. The bottom row plots the correlation between the maximum proportion of significant phase-locking values (see Supplemental Figure 3) and the maximum attentional phase shift, between 'Auditory-locked' and 'Visual-locked' analysis. Different colors represent different patients. Regression lines are plotted per patient and the black line is the regression for all data points across patients.

Taking corrected or uncorrected significant attentional phase shift did not change the results much. Adding all electrodes showing an entrainment skewed the distribution of attentional phase shifts towards small values, irrespective of the amount of phase locking and made the correlation not significant. This suggests that some electrodes, despite showing entrainment that was not due to a steady-state sensory response, were not influenced by attention.

Supplemental Figure 5. Location of electrodes tested by electrical stimulation.

a. All tested electrode from patients S3, S4, S5, S6, L4 and L5. Electrical stimulation at dark grey electrodes did not produce any deficit or sensation. Stimulation at colored electrodes did produce deficits in language task (red), reading task (pink), hand/arm motor task (green), tongue/mouth motor task (dark blue) or a sensation in the tongue/face (light blue). This map gives an idea of the error introduced by the co-registration and normalizations procedures, as well as displaying the electrodes. The location of the positive (colored) electrodes for different functions are reasonably separate but do not always respect well-known anatomo-functional boundaries such as the sylvian fissure or the central sulcus. Other factors for this include possible reorganization of the cortex in some patients due to epilepsy and distant stimulation by volume conduction. b. Location of all electrodes that showed a significant (uncorrected) entrainment in the 1.5Hz band. Light grey electrodes were not tested. Color coding is otherwise the same as in a. c. Location of all electrodes that showed a significant (corrected) attentional phase shift. Color coding is the same as in b. Circled electrodes are the ones described in Supplemental Figures 1 (red) and 2 (purple).

Supplemental Figure 6. Interictal activity at all implanted electrodes and comparison with significant electrodes

Location of all implanted electrodes displayed on a semi-transparent (a) or opaque (b) 3D reconstruction of the cortical surface of the MNI single subject. Electrodes identified as having frequent or occasional interictal epileptic activity after extensive review by clinicians of continuous recording from all electrodes, are displayed in orange and pink respectively. Electrodes showing significant attentional effects in our experiment are displayed in green. Note the prevalence of frequent interictal activity on depth electrodes implanted in hippocampal cortex, where no significant effects were found (in part because most of these electrodes were excluded from our analysis). Since electrodes were excluded from the analysis only on the basis of the data recorded during the experiment, some electrodes clinically identified as interictal, were included in our analysis. However, out of 72 significant electrodes, only 3 presented frequent interictal activity (in red) and 7 presented occasional interictal activity (in purple). On all analyzed electrodes, trials including interictal slowing or spikes were excluded form the analysis.

Supplemental Tables

Patient Recorded electrodes

Analyzed electrodes

Significantly entrained

% of analyzed

Significant phase shift

% of analyzed

% of entrained

S2 102 64 15 23% 4 6% 27% S3 96 75 20 27% 10 13% 50% S4 42 34 10 29% 9 26% 90% S5 99 80 13 16% 4 5% 31% S6 120 114 56 49% 12 11% 21% L4 126 112 45 40% 25 22% 56% L5 128 116 38 33% 8 7% 21%

Average 102 85 28 33% 10 12% 37% Total 713 595 197 33% 72 12% 37%

Supplemental Table 1: Number of implanted, analyzed and significant electrodes per patient

Patient Age Sex Age of onset Handedness Side of

implantation Language

lateralization S2 45 female 33 right bilateral left S3 44 female 6 right left Not tested S4 20 female 18 right left Not tested S5 22 male 15 right right Not tested S6 30 female 4 right left left L4 20 female 13 right left left L5 44 female 3 right/ambidextrous bilateral bilateral

Supplemental Table 2: Demographics and clinical information

Language lateralization was tested using the WADA test.

Supplemental Table 3: Behavioral performance.

Performance was better for visual targets than auditory targets: TR was faster, the percentage of correct hits larger and the proportion of hits on catch trials (wrong hits) smaller, but the difference was significant only for wrong hits (permutation test, p<10-4). There were no catch trials (targets when patients paid attention to another modality) for patient S2. Supplemental Table 2 gives the average performance for each patients in the auditory and visual tasks: there were no significant differences between visual and auditory RT’s and hit rates (mean RT: 535 ms vs. 562 ms, mean hit rate: 78% vs. 66%). However, there were significantly less responses to auditory targets in visual blocks than to visual targets in auditory blocks (7% vs. 13%, p<10-4, permutation test).

Auditory Attention Visual Attention

Patient Mean RT

Correct responses Wrong hits False

alarms Mean RT

Correct responses Wrong hits False

alarms S2 583 32/48 67% - - - 755 27/48 56% - - - S3 518 35/48 73% 14/48 29% 12 395 93/96 97% 18/96 19% 18 S4 552 47/64 73% 7/64 11% 15 526 52/64 81% 3/64 5% 19 S5 554 41/88 47% 13/88 15% 24 524 65/86 76% 6/89 7% 18 S6 522 22/40 55% 5/40 13% 3 485 32/40 80% 3/40 8% 14 L4 620 46/64 72% 4/64 6% 8 506 45/56 80% 2/56 4% 4 L5 584 37/48 77% 1/64 2% 11 557 35/48 73% 1/64 2% 11

Jitt

ered

ISI

Mean 562 66% 13% 535 78% 7%

S5 505 38/48 88% 1/48 2% 3 568 33/48 69% 3/48 6% 9

S6 513 28/40 70% 4/40 10% 1 582 28/40 70% 1/40 3% 9

Reg

ular

ISI

Mean 509 79% 6% 575 70% 5%

Patient Neuropsychological testing

Symptoms Imaging Monitoring Implantation Resection

S2 WADA showed left language dominance with memory 7/8 on Right injection and 6/6 on Left injection.

Seizure onset was at age 33 with episodes of staring, unresponsiveness and oral automatisms. She reports at times an aura of dizziness and a “shaky feeling.” She is amnesic for most events. She has occasional secondarily generalized seizures, and seizures are worse around the time of her menses.

MRI: no significant abnormalities; PET: showed right mesial and inferior temporal hypometabolism.

Scalp: interictal bilateral temporal IED’s, right greater than left temporal delta slowing, and right TIRDA. 3 seizures were captured, onset with left temporal delta slowing which quickly settled into right temporal rhythmic activity up to 8 Hz. MEG: right inferior frontal and anterior temporal epileptiform disturbances

bilateral subtemporal, supra-temporal, parietal and frontal strips, temporal depth.

Right anteromedial temporal lobectomy with hippocampal sclerosis on path.

S3 Non-lateralizing or localizing.

Semiology consists of behavioral arrest, eye fluttering and dysphasia, with some word finding difficulties interictally. medical history and imaging most consistent with partial left MCA infarction at age 6years. Now with medically intractable epilepsy consisting of frequent partial seizures with sensory phenomenon over right side of head and face often with some degree of expressive and receptive language impairment. Recent episode of complex partial status epilepticus in Fall of 2007 when medications were adjusted.

MRI: focal atrophy and encephalomalacia involving the left temporal horn, left inferior frontal region and left caudate head PET: decreased metabolism in L hemisphere, greatest over left temporal region.

Scalp: Int. left temporal slowing and frequent broad left temporal sharp waves. Innumerable probable SPS without EEG correlate. 2 CPS with left temporal onset

left subtemporal strips and temporo-parieto-frontal grid

No

L4 Executive and language dysfunctions suggestive of mild left frontal impairment. WADA: Left language; bilateral memory.

Episodes since age 13. Consist of staring spells that last for seconds, followed by a period of confusion for a few minutes. Often speaks nonsensical sentences during seizures.

Hyperintensity on the T2 and FLAIR in the left thalamus.

Scalp: features of both focal and generalized epilepsy. Both left and right onsets. Intra: Occipito-temporal onsets, more pronounced on the left Seizures at multiple sites.

Left parieto-temporo-frontal grid, strips all over the right and left hemispheres. Left hippocampal depths.

No.

L5 WADA: bilateral language. Bilateral memory

Episodes since age 3 after measles infection Auras of abdominal rising and a funny pleasant feeling Often followed by right hand shaking and right facial twitching and then ad post-ictal aphasia and alexia

MRI: Left temporal atrophy, mesial temporal sclerosis Multiple high signals on T2 that are consistent with her history of childhood encephalitis PET: left parieto-temporal hypometabolism and right temporal hypometabolism

Scalp: left temporal onsets right temporal interictal spikes

Left parieto-temporo-frontal grid, strips all over the right and left hemispheres. Bilateral hippocampal depths.

Left anterior temporal with 2/3 of the hippocampus.

S4 history of left parietal AVM s/p hemorrhage and partial resection. Since the surgery she has gone on to develop medically refractory Localization Related Epilepsy. Her seizures

MRI: recurrent/residual left frontal/parietal AVM.

Scalp: In the EMU one GTC was captured with onset from the Left parietal region. Multiple simple partial seizures were captured without

left parietal grid.

No

usually begin with an aura of a rushing feeling in her chest accompanied by anxiety and tingling/pressure in her right arm and hand. These may then progress to confusion, body stiffening and at times generalized tonic-clonic activity.

clear EEG correlate.

S5 suggestion of right temporal and frontal dysfunction.

since age 15. Current semiology on medications is characterized by a blank stare noted by his parents, trying to stand up if seated and then consistent head deviation to the left within seconds and subsequent generalized body stiffening though he does not always fall. The seizures last 10-15 seconds and he does not bite his tongue or urinate on himself. They typically occur in the morning. No evidence of seizures has been noted overnight. Patient says he has no warning and no memory for the events. He typically has a post-ictal headache and can sleep for the next 24 hours.

MRI: non lesional Interictal PET: hypometabolism in right temporal lobe.

MEG: definite epileptogenic region in the anterior frontal lobe; slow waves seen were seen posterior frontally with possible extension into the parietal lobe.

right parieto-frontal grid.

No

S6 Left temporal dysfunction (language). Possible left frontal dysfunction (executive) WADA: left language, bilateral memory.

Her first seizure occurred at age 4 with a fever, when she “passed out” with her eyes rolled back. She was treated with Phenobarbital for 2.5 years, and was then seizure free until seizures recurred at age 16, after which time they have been medically refractory. She current has complex partial seizures with unclear frequency, with decreased awareness +/- behavioral arrest, oral and/or manual automatisms, and drooling, at times progressing to twisting of her body and head to the right. There are no known post-ictal focal deficits. She is often unaware of these seizures, but has been seen to have them at least 1-2 times/month. She has also described “bigger seizures,” with tonic and clonic movements, occurring without warning (seen on monitoring to include forced right head deviation, right hand clonic movements and left arm tonic movements with secondary generalization).

MRI: Left > right temporal-occipital heterotopia - Atrophy of the left temporal lobe; PET: Decreased uptake in left medial and anterior temporal regions; SPECT: Interictal: decreased Left mesiotemporal region - Ictal: increased Left mesiotemporal region.

EEG Interictal: left temporal slowing; Epileptiform discharges: Fp1, F3, F7, P7. Ictal - Left hemisphere and F3, F7

left inferior and middle temporal gyri, sparing areas potentially important for language function (particularly G37, G38, G45 and G46, as well as G35-G42).

Supplemental Table 4: Patient Clinical information