Cerebral Cortex Advance Access published June 3, 2015 Cerebral Cortex, 2015, 1–10 doi: 10.1093/cercor/bhv128 Original Article
ORIGINAL ARTICLE
Ketamine Alters Outcome-Related Local Field Potentials in Monkey Prefrontal Cortex 1
Department of Anatomy and Cell Biology, 2Department of Physiology and Pharmacology, 3Department of Psychology, Western University, London, ON, Canada, 4Department of Biology, York University, Toronto, ON, Canada and 5Robarts Research Institute, London, ON, Canada Address correspondence to Dr Stefan Everling, Centre for Functional and Metabolic Mapping, Robarts Research Institute, 1151 Richmond Street North, London, ON, Canada N6A 5B7. Email: [email protected]
Abstract A subanesthetic dose of the noncompetitive N-methyl--aspartate receptor antagonist ketamine is known to induce a schizophrenia-like phenotype in humans and nonhuman primates alike. The transient behavioral changes mimic the positive, negative, and cognitive symptoms of the disease but the neural mechanisms behind these changes are poorly understood. A growing body of evidence indicates that the cognitive control processes associated with prefrontal cortex (PFC) regions relies on groups of neurons synchronizing at narrow-band frequencies measurable in the local field potential (LFP). Here, we recorded LFPs from the caudo-lateral PFC of 2 macaque monkeys performing an antisaccade task, which requires the suppression of an automatic saccade toward a stimulus and the initiation of a goal-directed saccade in the opposite direction. Preketamine injection activity showed significant differences in a narrow 20–30 Hz beta frequency band between correct and error trials in the postsaccade response epoch. Ketamine significantly impaired the animals’ performance and was associated with a loss of the differences in outcome-specific beta-band power. Instead, we observed a large increase in high-gamma-band activity. Our results suggest that the PFC employs beta-band synchronization to prepare for top–down cognitive control of saccades and the monitoring of task outcome. Key words: beta-band, gamma-band, ketamine, local field potential, performance monitoring, prefrontal cortex
Introduction Schizophrenia is a debilitating neuropsychiatric disease affecting nearly 1% of the world’s population, yet its underlying neural mechanisms remain poorly understood. An old theory (Wernicke 1906) that has gained traction in recent years is the notion that schizophrenia is not caused by focal brain alterations but by a dysconnectivity in communication between brain regions (Friston 1998; Phillips and Silverstein 2003; Stephan et al. 2006). Recent functional connectivity studies have shown that the acute phases of the illness are associated with dysconnectivity of the prefrontal cortex (PFC) (Anticevic et al. 2015), a key region for cognitive function in primates (Miller and Cohen 2001). Similar changes in functional connectivity were observed in healthy volunteers after subanesthetics doses of the noncompetitive
N-methyl--aspartate (NMDA) glutamate receptor antagonist, ketamine, which also induces a transient schizophrenia-like behavioral phenotype (Krystal et al. 1994; Adler et al. 1999). Findings from electroencephalogram (EEG) recordings in schizophrenia patients support the hypothesis of a dysconnectivity in communication between brain regions, which is known to be mediated by neural oscillations (Salinas and Sejnowski 2000, 2001; Tiesinga and Sejnowski 2004; Womelsdorf et al. 2014). The most prominent findings are an increase in gammaband EEG activity (>30 Hz) (Barr et al. 2010; Sun et al. 2011) and a decrease in beta-band EEG activity (15–30 Hz) (Krishnan et al. 2005; Uhlhaas et al. 2006; Hirano et al. 2008). While EEG recordings in nonhuman primates have shown that subanaesthetics doses of ketamine reduce mismatch-negative
© The Author 2015. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected]
1
Downloaded from http://cercor.oxfordjournals.org/ at Simon Fraser University on June 5, 2015
Kevin J. Skoblenick1, Thilo Womelsdorf4, and Stefan Everling1,2,3,5
2
| Cerebral Cortex
and P3a event-related potentials (Gil-da-Costa et al. 2013), it is unknown whether ketamine also alters oscillatory potentials in the PFC. Here we investigated the effects of a subanesthetic dose of ketamine on outcome-related local field potentials (LFPs) in the macaque PFC during an antisaccade task, which requires subjects to suppress a saccade towards a flashed peripheral stimulus instead to generate a saccade to the opposite direction (Munoz and Everling 2004). Poor antisaccade task performance is a potential biomarker for schizophrenia (Crawford et al. 1998; Hutton and Ettinger 2006) and has also been found in nonhuman primates following ketamine administration (Condy et al. 2005; Skoblenick and Everling 2012, 2014).
Materials and Methods Animals
A
Fixation 100 ms
Rule cue 1000–1200 ms
Stimulus onset 500 ms
Both animals were trained to perform the antisaccade task to receive a liquid reward, as described previously by our laboratory (Johnston and Everling 2011; Skoblenick and Everling 2012). Briefly, the animal was enclosed in a light and sound-attenuated chamber with their head restrained and oriented toward a computer monitor displaying the task. Each trial began with a fixation stimulus, which was replaced with a rule cue after fixation for 100 ms. Monkey O was trained that a green-colored rule cue indicated a prosaccade task and a red-colored cue indicated an antisaccade task. The color rules were reversed for training Monkey T. After a randomized 1000–1200 ms instruction period, a white, circular, peripheral stimulus appeared either 8° to the left or right of the central fixation cue. Prosaccade trials required the animal look at this stimulus, while antisaccade trials required the animal to look toward a blank location on the screen diametrically opposite to the stimulus’ position. Following correct trials, a liquid reward was delivered to the monkey’s mouth through a sipping tube immediate after a saccade to the correct target location. Eye gazes in any direction other than toward or away from the stimuli terminated the trial immediately. Following the completion of a trial, the screen was blanked and a new trial began after a 500-ms intertrial interval (Fig. 1A). Trials were presented in a randomized order such that pro- and antisaccade trials occurred equally as often. Owing to the minimal amount of erroneous prosaccade trials in the preketamine condition, analyses between correct and error trials were performed exclusively on the antisaccade condition. The animal’s eye position was recorded and digitized at either 120 Hz using an ISCAN infrared pupillary tracking system for Monkey O (ISCAN, Woburn, MA, USA) or at 1000 Hz using an Eyelink 1000 infrared pupillary tracking system for both Monkey O and Monkey T (SR Research, Mississauga, ON, Canada).
Recording Pro
Anti
B
Monkey O (left)
Monkey O (right)
Monkey T (left)
AS
AS
AS
P PS
PS
L
M A
PS 1 cm
Figure 1. Task and recording locations. (A) Organization of antisaccade task. Monkey O was trained with a green rule cue indicating a prosaccade trial and a red rule cue indicating an antisaccade trial. The colors were inversed for Monkey T. After a correct saccade, the animal was rewarded and the screen was blanked for 500 ms before the fixation cue appeared again. (B) Reconstruction of
The setup for multielectrode recording sessions began a day before the experimental session with the installation of the semichronic multielectrode grid (Neuronitek, London, ON, Canada) into the recording chamber. The 36 tungsten electrodes (FHC, Bowdoin, ME, USA) were lowered through a silicone membrane and into the monkey’s lateral PFC, providing 32 recording channels and 4 reference channels. Each electrode was lowered manually using a micro-screwdriver until background activity was observed on a maximal number of recording channels. The animal was returned to its cage until the next day so that the electrodes had time to settle in the cortical tissue. On an experimental session day, the animal was returned to the sound-attenuated experimental chamber and the electrode grid was connected to a head stage and amplifying unit. Neuronal spiking activity and LFP activity from each channel were combined with performance and eye-tracking data in a multiacquisition processor (MAP) system (Plexon, Dallas, TX, USA) and sorted offline using 2D and 3D principal component analysis. Subsequent recording days only required reconnecting the head stage and a minimal adjustment to each electrode’s depth before spiking neuronal activity was observed. The multielectrode grid was left implanted for 2 weeks, after which it was removed for cleaning and sterilization of the recording chamber. The LFP data were recorded simultaneously with single unit activity that was analyzed and reported separately (Skoblenick and Everling 2014).
chamber placements. Monkey O had bilateral recording chamber implantation and had 2 sessions recorded from the right hemisphere and 3 sessions recorded from the left hemisphere. Monkey T had all 3 sessions recorded from his
Drug Administration
unilateral left hemisphere chamber. as, arcuate sulcus; ps, principal sulcus; P,
Each session began with a 15-min preinjection period, during which the animal performed 200–250 trials with accompanying
posterior; A, anterior; L, lateral; M, medial.
Downloaded from http://cercor.oxfordjournals.org/ at Simon Fraser University on June 5, 2015
The experiments were performed in accordance with the Canadian Council of Animal Care policy on the use of laboratory animals and all procedures were approved by the Animal Use Subcommittee of the University Western Ontario Council on Animal Care. For this study, 2 male rhesus monkeys (Macaca mulatta) weighing 5 kg (Monkey T) and 7 kg (Monkey O) performed the behavioral tasks. The animals were implanted with a recording chamber located above their lateral PFC and a plastic head restraint, as previously described (Johnston and Everling 2006). Postsurgical treatments included analgesics, prophylactic antibiotics, and oversight by the university veterinarian. Following surgical recovery, animals had cranial MR imaging to obtain anatomical localization of the recording chambers (Fig. 1B).
Behavioral Task
Effect of Ketamine on Local Field Potential in PFC
Data Analysis Electrophysiological data were analyzed using custom scripts for Matlab (Mathworks) that made use of the FieldTrip toolbox (http://fieldtrip.fcdonders.nl/) developed at the Donders Institute for Brain, Cognition and Behaviour (Oostenveld et al. 2011). For LFP analysis, the continuous analog signal was divided into discrete trials using event time markers provided by the Plexon MAP system. Data were filtered with a low-pass filter at 150 Hz and line noise was removed at 60 and 120 Hz using a discrete Fourier transform (see (Womelsdorf et al. 2006)). Z-score thresholding, and independent component analysis were used to detect and discard any additional mechanical artifacts in the analog signal. To remove the reward artifact, the data were first downsampled to 1/100th the sampling frequency after which component analysis was run to manually identify those components that contained the artifacts. The component analysis was then reperformed on the original data without downsampling to remove the artifact components from the final dataset (Supplementary Fig. 1). To determine time-locked LFP power, frequency analysis was performed using the multitaper method with a discrete prolate spheroidal sequence taper set around a 0.667-s window every 50 ms for the low-frequency range (1–60 Hz) and a 0.33-s window every 50 ms for the high-frequency range (40–150 Hz). The LFP data for each trial were normalized to the oscillatory activity on the same channel during the preceding intertrial interval (500-ms preceding fixation onset) resulting in a Z-score that could be compared between channels, sessions, and animals. There was no significant difference between the pre- and postketamine baseline power spectrums (Supplementary Fig. 2). Statistical analysis on the resulting time–frequency–LFP power maps used a nonparametric cluster-based analysis that created a T-value map for the significance level between 2 conditions and highlighted time–frequency epochs with statistical significance and corrected for multiple comparisons across time and frequency bins (Maris and Oostenveld 2007; Womelsdorf et al. 2010). Frequency data were divided into 0.33-Hz bins, while time data were measure in 0.1-s bins. The bootstrapping analyses were performed against 10 000 permutations of the data to create the significance maps. Neuronal spiking activity was analyzed as previously described (Skoblenick and Everling 2012) and analyses into the signal-to-noise ratio differences between correct and
| 3
error conditions are previously discussed in Skoblenick and Everling (2014).
Results Data were collected in the dorsolateral PFC (Fig. 1B) during 8 testing sessions (5 with Monkey O and 3 with Monkey T) yielding 158 LFP channels and 215 neurons for analysis. Channels without usable signals were removed from the analysis. Behavioral effects for this dataset have been described previously (Skoblenick and Everling 2014). We limited the analysis to antisaccade trials, as the animals made no or very few errors on prosaccade trials before ketamine as previously reported (Skoblenick and Everling 2014).
Prefrontal Cortex Exhibits Outcome-Dependent Beta-Band LFP Activity To examine how the beta-band range of LFP activity may be involved in a task requiring explicit cognitive control, we recorded the LFP signal in 2 monkeys. After lower frequency analysis was performed (1–60 Hz), a clear time–frequency epoch emerged with band-limited power modulation following the animal’s saccade response to the trial stimulus. The response occurred most strongly between 15 and 30 Hz, corresponding to beta-band oscillations. Furthermore, this response appeared to be outcome-specific, showing a stronger effect following trials in which the animal made a correct response (Fig. 2). After the effect was visualized the statistical strength of these findings were tested with a cluster-based analysis (see Materials and Methods). The resulting T-score map highlighted a cluster between 0.3 and 0.75 s after saccade onset that showed a significant difference (P < 0.05, multiple comparison corrected) in the LFP activation between correct and error trials. To better illustrate the evolution of this difference, the mean LFP power of the relevant oscillatory range (15–30 Hz) was plotted as mean ± SEM over the course of the trial (Fig. 2B).
Ketamine Reduces Performance Selectivity of Beta-Band LFP Activity After observing the selectivity in the beta-band response following task completion, we next analyzed how these changes were affected by the administration of a subanesthetic dose of ketamine. Figure 3 shows that ketamine decreased performance selectivity in the beta-band such that no significant difference was found between correct and error trials in the beta-band (Fig. 3A). The overall beta-band activity was decreased ( preketamine mean Z-score: 0.132 ± 0.267, postketamine mean Z-score: −0.278 ± 0.205, P < 0.001) to the detriment of the correct-selective epoch beginning 200 ms after saccade onset. A nonsignificant increase in activity in the 40-Hz range was also noted after ketamine administration, and this 40 Hz activity showed no selectivity between correct and error trials. The differences in (correct – error) LFP power were compared for statistical significance and the cluster-based analysis once again found a time–frequency epoch significantly different (P < 0.05, multiple comparison corrected) between pre- and postketamine conditions (Fig. 3B). The evolution of this change in selectivity over the course of the trial was also plotted as mean power (15–30 Hz) ± SEM to illustrate how pre- and postketamine conditions differed (Fig. 3C). Power spectrograms also highlight the beta-band selectivity before ketamine administration and the changes
Downloaded from http://cercor.oxfordjournals.org/ at Simon Fraser University on June 5, 2015
PFC recordings. The animals both made on average
ORIGINAL ARTICLE
Ketamine Alters Outcome-Related Local Field Potentials in Monkey Prefrontal Cortex 1
Department of Anatomy and Cell Biology, 2Department of Physiology and Pharmacology, 3Department of Psychology, Western University, London, ON, Canada, 4Department of Biology, York University, Toronto, ON, Canada and 5Robarts Research Institute, London, ON, Canada Address correspondence to Dr Stefan Everling, Centre for Functional and Metabolic Mapping, Robarts Research Institute, 1151 Richmond Street North, London, ON, Canada N6A 5B7. Email: [email protected]
Abstract A subanesthetic dose of the noncompetitive N-methyl--aspartate receptor antagonist ketamine is known to induce a schizophrenia-like phenotype in humans and nonhuman primates alike. The transient behavioral changes mimic the positive, negative, and cognitive symptoms of the disease but the neural mechanisms behind these changes are poorly understood. A growing body of evidence indicates that the cognitive control processes associated with prefrontal cortex (PFC) regions relies on groups of neurons synchronizing at narrow-band frequencies measurable in the local field potential (LFP). Here, we recorded LFPs from the caudo-lateral PFC of 2 macaque monkeys performing an antisaccade task, which requires the suppression of an automatic saccade toward a stimulus and the initiation of a goal-directed saccade in the opposite direction. Preketamine injection activity showed significant differences in a narrow 20–30 Hz beta frequency band between correct and error trials in the postsaccade response epoch. Ketamine significantly impaired the animals’ performance and was associated with a loss of the differences in outcome-specific beta-band power. Instead, we observed a large increase in high-gamma-band activity. Our results suggest that the PFC employs beta-band synchronization to prepare for top–down cognitive control of saccades and the monitoring of task outcome. Key words: beta-band, gamma-band, ketamine, local field potential, performance monitoring, prefrontal cortex
Introduction Schizophrenia is a debilitating neuropsychiatric disease affecting nearly 1% of the world’s population, yet its underlying neural mechanisms remain poorly understood. An old theory (Wernicke 1906) that has gained traction in recent years is the notion that schizophrenia is not caused by focal brain alterations but by a dysconnectivity in communication between brain regions (Friston 1998; Phillips and Silverstein 2003; Stephan et al. 2006). Recent functional connectivity studies have shown that the acute phases of the illness are associated with dysconnectivity of the prefrontal cortex (PFC) (Anticevic et al. 2015), a key region for cognitive function in primates (Miller and Cohen 2001). Similar changes in functional connectivity were observed in healthy volunteers after subanesthetics doses of the noncompetitive
N-methyl--aspartate (NMDA) glutamate receptor antagonist, ketamine, which also induces a transient schizophrenia-like behavioral phenotype (Krystal et al. 1994; Adler et al. 1999). Findings from electroencephalogram (EEG) recordings in schizophrenia patients support the hypothesis of a dysconnectivity in communication between brain regions, which is known to be mediated by neural oscillations (Salinas and Sejnowski 2000, 2001; Tiesinga and Sejnowski 2004; Womelsdorf et al. 2014). The most prominent findings are an increase in gammaband EEG activity (>30 Hz) (Barr et al. 2010; Sun et al. 2011) and a decrease in beta-band EEG activity (15–30 Hz) (Krishnan et al. 2005; Uhlhaas et al. 2006; Hirano et al. 2008). While EEG recordings in nonhuman primates have shown that subanaesthetics doses of ketamine reduce mismatch-negative
© The Author 2015. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: [email protected]
1
Downloaded from http://cercor.oxfordjournals.org/ at Simon Fraser University on June 5, 2015
Kevin J. Skoblenick1, Thilo Womelsdorf4, and Stefan Everling1,2,3,5
2
| Cerebral Cortex
and P3a event-related potentials (Gil-da-Costa et al. 2013), it is unknown whether ketamine also alters oscillatory potentials in the PFC. Here we investigated the effects of a subanesthetic dose of ketamine on outcome-related local field potentials (LFPs) in the macaque PFC during an antisaccade task, which requires subjects to suppress a saccade towards a flashed peripheral stimulus instead to generate a saccade to the opposite direction (Munoz and Everling 2004). Poor antisaccade task performance is a potential biomarker for schizophrenia (Crawford et al. 1998; Hutton and Ettinger 2006) and has also been found in nonhuman primates following ketamine administration (Condy et al. 2005; Skoblenick and Everling 2012, 2014).
Materials and Methods Animals
A
Fixation 100 ms
Rule cue 1000–1200 ms
Stimulus onset 500 ms
Both animals were trained to perform the antisaccade task to receive a liquid reward, as described previously by our laboratory (Johnston and Everling 2011; Skoblenick and Everling 2012). Briefly, the animal was enclosed in a light and sound-attenuated chamber with their head restrained and oriented toward a computer monitor displaying the task. Each trial began with a fixation stimulus, which was replaced with a rule cue after fixation for 100 ms. Monkey O was trained that a green-colored rule cue indicated a prosaccade task and a red-colored cue indicated an antisaccade task. The color rules were reversed for training Monkey T. After a randomized 1000–1200 ms instruction period, a white, circular, peripheral stimulus appeared either 8° to the left or right of the central fixation cue. Prosaccade trials required the animal look at this stimulus, while antisaccade trials required the animal to look toward a blank location on the screen diametrically opposite to the stimulus’ position. Following correct trials, a liquid reward was delivered to the monkey’s mouth through a sipping tube immediate after a saccade to the correct target location. Eye gazes in any direction other than toward or away from the stimuli terminated the trial immediately. Following the completion of a trial, the screen was blanked and a new trial began after a 500-ms intertrial interval (Fig. 1A). Trials were presented in a randomized order such that pro- and antisaccade trials occurred equally as often. Owing to the minimal amount of erroneous prosaccade trials in the preketamine condition, analyses between correct and error trials were performed exclusively on the antisaccade condition. The animal’s eye position was recorded and digitized at either 120 Hz using an ISCAN infrared pupillary tracking system for Monkey O (ISCAN, Woburn, MA, USA) or at 1000 Hz using an Eyelink 1000 infrared pupillary tracking system for both Monkey O and Monkey T (SR Research, Mississauga, ON, Canada).
Recording Pro
Anti
B
Monkey O (left)
Monkey O (right)
Monkey T (left)
AS
AS
AS
P PS
PS
L
M A
PS 1 cm
Figure 1. Task and recording locations. (A) Organization of antisaccade task. Monkey O was trained with a green rule cue indicating a prosaccade trial and a red rule cue indicating an antisaccade trial. The colors were inversed for Monkey T. After a correct saccade, the animal was rewarded and the screen was blanked for 500 ms before the fixation cue appeared again. (B) Reconstruction of
The setup for multielectrode recording sessions began a day before the experimental session with the installation of the semichronic multielectrode grid (Neuronitek, London, ON, Canada) into the recording chamber. The 36 tungsten electrodes (FHC, Bowdoin, ME, USA) were lowered through a silicone membrane and into the monkey’s lateral PFC, providing 32 recording channels and 4 reference channels. Each electrode was lowered manually using a micro-screwdriver until background activity was observed on a maximal number of recording channels. The animal was returned to its cage until the next day so that the electrodes had time to settle in the cortical tissue. On an experimental session day, the animal was returned to the sound-attenuated experimental chamber and the electrode grid was connected to a head stage and amplifying unit. Neuronal spiking activity and LFP activity from each channel were combined with performance and eye-tracking data in a multiacquisition processor (MAP) system (Plexon, Dallas, TX, USA) and sorted offline using 2D and 3D principal component analysis. Subsequent recording days only required reconnecting the head stage and a minimal adjustment to each electrode’s depth before spiking neuronal activity was observed. The multielectrode grid was left implanted for 2 weeks, after which it was removed for cleaning and sterilization of the recording chamber. The LFP data were recorded simultaneously with single unit activity that was analyzed and reported separately (Skoblenick and Everling 2014).
chamber placements. Monkey O had bilateral recording chamber implantation and had 2 sessions recorded from the right hemisphere and 3 sessions recorded from the left hemisphere. Monkey T had all 3 sessions recorded from his
Drug Administration
unilateral left hemisphere chamber. as, arcuate sulcus; ps, principal sulcus; P,
Each session began with a 15-min preinjection period, during which the animal performed 200–250 trials with accompanying
posterior; A, anterior; L, lateral; M, medial.
Downloaded from http://cercor.oxfordjournals.org/ at Simon Fraser University on June 5, 2015
The experiments were performed in accordance with the Canadian Council of Animal Care policy on the use of laboratory animals and all procedures were approved by the Animal Use Subcommittee of the University Western Ontario Council on Animal Care. For this study, 2 male rhesus monkeys (Macaca mulatta) weighing 5 kg (Monkey T) and 7 kg (Monkey O) performed the behavioral tasks. The animals were implanted with a recording chamber located above their lateral PFC and a plastic head restraint, as previously described (Johnston and Everling 2006). Postsurgical treatments included analgesics, prophylactic antibiotics, and oversight by the university veterinarian. Following surgical recovery, animals had cranial MR imaging to obtain anatomical localization of the recording chambers (Fig. 1B).
Behavioral Task
Effect of Ketamine on Local Field Potential in PFC
Data Analysis Electrophysiological data were analyzed using custom scripts for Matlab (Mathworks) that made use of the FieldTrip toolbox (http://fieldtrip.fcdonders.nl/) developed at the Donders Institute for Brain, Cognition and Behaviour (Oostenveld et al. 2011). For LFP analysis, the continuous analog signal was divided into discrete trials using event time markers provided by the Plexon MAP system. Data were filtered with a low-pass filter at 150 Hz and line noise was removed at 60 and 120 Hz using a discrete Fourier transform (see (Womelsdorf et al. 2006)). Z-score thresholding, and independent component analysis were used to detect and discard any additional mechanical artifacts in the analog signal. To remove the reward artifact, the data were first downsampled to 1/100th the sampling frequency after which component analysis was run to manually identify those components that contained the artifacts. The component analysis was then reperformed on the original data without downsampling to remove the artifact components from the final dataset (Supplementary Fig. 1). To determine time-locked LFP power, frequency analysis was performed using the multitaper method with a discrete prolate spheroidal sequence taper set around a 0.667-s window every 50 ms for the low-frequency range (1–60 Hz) and a 0.33-s window every 50 ms for the high-frequency range (40–150 Hz). The LFP data for each trial were normalized to the oscillatory activity on the same channel during the preceding intertrial interval (500-ms preceding fixation onset) resulting in a Z-score that could be compared between channels, sessions, and animals. There was no significant difference between the pre- and postketamine baseline power spectrums (Supplementary Fig. 2). Statistical analysis on the resulting time–frequency–LFP power maps used a nonparametric cluster-based analysis that created a T-value map for the significance level between 2 conditions and highlighted time–frequency epochs with statistical significance and corrected for multiple comparisons across time and frequency bins (Maris and Oostenveld 2007; Womelsdorf et al. 2010). Frequency data were divided into 0.33-Hz bins, while time data were measure in 0.1-s bins. The bootstrapping analyses were performed against 10 000 permutations of the data to create the significance maps. Neuronal spiking activity was analyzed as previously described (Skoblenick and Everling 2012) and analyses into the signal-to-noise ratio differences between correct and
| 3
error conditions are previously discussed in Skoblenick and Everling (2014).
Results Data were collected in the dorsolateral PFC (Fig. 1B) during 8 testing sessions (5 with Monkey O and 3 with Monkey T) yielding 158 LFP channels and 215 neurons for analysis. Channels without usable signals were removed from the analysis. Behavioral effects for this dataset have been described previously (Skoblenick and Everling 2014). We limited the analysis to antisaccade trials, as the animals made no or very few errors on prosaccade trials before ketamine as previously reported (Skoblenick and Everling 2014).
Prefrontal Cortex Exhibits Outcome-Dependent Beta-Band LFP Activity To examine how the beta-band range of LFP activity may be involved in a task requiring explicit cognitive control, we recorded the LFP signal in 2 monkeys. After lower frequency analysis was performed (1–60 Hz), a clear time–frequency epoch emerged with band-limited power modulation following the animal’s saccade response to the trial stimulus. The response occurred most strongly between 15 and 30 Hz, corresponding to beta-band oscillations. Furthermore, this response appeared to be outcome-specific, showing a stronger effect following trials in which the animal made a correct response (Fig. 2). After the effect was visualized the statistical strength of these findings were tested with a cluster-based analysis (see Materials and Methods). The resulting T-score map highlighted a cluster between 0.3 and 0.75 s after saccade onset that showed a significant difference (P < 0.05, multiple comparison corrected) in the LFP activation between correct and error trials. To better illustrate the evolution of this difference, the mean LFP power of the relevant oscillatory range (15–30 Hz) was plotted as mean ± SEM over the course of the trial (Fig. 2B).
Ketamine Reduces Performance Selectivity of Beta-Band LFP Activity After observing the selectivity in the beta-band response following task completion, we next analyzed how these changes were affected by the administration of a subanesthetic dose of ketamine. Figure 3 shows that ketamine decreased performance selectivity in the beta-band such that no significant difference was found between correct and error trials in the beta-band (Fig. 3A). The overall beta-band activity was decreased ( preketamine mean Z-score: 0.132 ± 0.267, postketamine mean Z-score: −0.278 ± 0.205, P < 0.001) to the detriment of the correct-selective epoch beginning 200 ms after saccade onset. A nonsignificant increase in activity in the 40-Hz range was also noted after ketamine administration, and this 40 Hz activity showed no selectivity between correct and error trials. The differences in (correct – error) LFP power were compared for statistical significance and the cluster-based analysis once again found a time–frequency epoch significantly different (P < 0.05, multiple comparison corrected) between pre- and postketamine conditions (Fig. 3B). The evolution of this change in selectivity over the course of the trial was also plotted as mean power (15–30 Hz) ± SEM to illustrate how pre- and postketamine conditions differed (Fig. 3C). Power spectrograms also highlight the beta-band selectivity before ketamine administration and the changes
Downloaded from http://cercor.oxfordjournals.org/ at Simon Fraser University on June 5, 2015
PFC recordings. The animals both made on average
