PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Science. Author manuscript; available in PMC Jun 10, 2012.
Published in final edited form as:
PMCID: PMC3371378
NIHMSID: NIHMS378577
Selective Attention from Voluntary Control of Neurons in Prefrontal Cortex
Robert J. Schafer1 and Tirin Moore1,2*
1Department of Neurobiology, Stanford University School of Medicine, Stanford, CA 94305, USA
2Howard Hughes Medical Institute, Stanford University School of Medicine, Stanford, CA 94305, USA
*To whom correspondence should be addressed. tirin/at/stanford.edu
Present address: McGovern Institute for Brain Research, Massachusetts Institute of Technology, Cambridge, MA 02139, USA.
Animals can learn to voluntarily control neuronal activity within various brain areas through operant conditioning, but the relevance of that control to cognitive functions is unknown. We show that monkeys can control the activity of neurons within the frontal eye field (FEF), an oculomotor area of prefrontal cortex. However, operantly driven FEF activity was primarily associated with selective visual attention, and not oculomotor preparation. Attentional effects were untrained, and were observed both behaviorally and neurophysiologically. Furthermore, selective attention correlated with voluntary, but not spontaneous, fluctuations in FEF activity. Our results reveal a specific association of voluntarily driven neuronal activity with “top-down” attention and suggest a basis for the use of neurofeedback training to treat disorders of attention.
Animal and human subjects can learn to alter their own brain activity when provided feedback (14). Voluntary control of neuronal activity is likely associated with changes in behavior or cognitive functions, but that relationship is unclear. Operant control of motor cortical neurons is typically dissociated from movement production (58), and there are no clear behavioral consequences of operant control of neurons in other brain structures (1, 4). Naturally, one might ask whether a chosen control strategy can elicit untrained behavioral or neurophysiological outcomes. To address this question, we examined the consequences of voluntary control of neurons in the frontal eye field (FEF), a visuomotor area within prefrontal cortex with a known role in the programming of saccadic eye movements (9) and visual spatial attention (10) (Fig. 1A).
Fig. 1
Fig. 1
Operant control paradigm. (A) Location of the FEF in the arcuate sulcus (shading) shown in a lateral view of a monkey brain (top). Eye traces at bottom-left show saccades evoked with 50μA microstimulation of an FEF site. Bottom-middle and right (more ...)
We first asked whether the activity of FEF neurons can be controlled voluntarily by the monkey without explicit training on any task. We used an operant control paradigm (2) in which the monkey received rewards for alternately increasing and decreasing the firing rates (FRs) of FEF neurons during fixation (11) (Fig. 1B,C) [supporting online materials (SOM)]. During trials the monkey received auditory feedback via pure tones, the pitch of which corresponded to the instantaneous FR of multiunit activity (MUA) at an FEF recording site. In blocks of “Up” trials, the monkey received a reward each time the FR, and thus the tone pitch, reached a high threshold (Fig. 1C). In blocks of “Down” trials, the monkey was rewarded each time a low threshold was reached.
We recorded from 94 FEF sites in two monkeys (Monkey B: 49, Monkey C: 45) during operant control. Figure 2A–C shows the results of a representative experiment. At this site, the MUA FR on Up trials was greater than on Down trials (Fig. 2A; P < 10−4), indicating that the monkey modulated the MUA in the rewarded direction. This FR difference persisted through the entire trial, and was maintained between blocks of Up and Down trials (Fig. 2B). We quantified neuronal control with a “control index” (CI), with positive CIs indicating changes in activity in the rewarded direction. The overall CI for the example site was 0.064, corresponding to a 13.7% increase in FR from Down to Up trials (Fig. 2C).
Fig. 2
Fig. 2
Voluntary control of FEF neurons. (A) Average MUA over the course of Up (red) and Down (blue) trials during an example experiment. (B) Firing rates for each Down and Up trial in A are indicated by blue and red triangles, respectively, with the mean firing (more ...)
Across all experiments, monkeys exerted modest, but reliable, control over the FRs of FEF neurons. The mean CI across sites was 0.031 (Fig. 2D; Monkey B: P = 0.0086, Monkey C: P = 0.0010; P = 10−4), corresponding to a 6.4% difference in FR. However, the magnitude of control varied considerably throughout the course of each experiment (P = 0.0045; Fig. S1). Furthermore, for individual FEF sites we found both significant positive and negative effects of control (SOM). Fifty-five experiments (59%) showed individually significant effects of control, and of these, 38 (69%) had positive CIs, with a mean of 0.080 (P < 10−4), and 17 (31%) had negative CIs, with a mean of −0.018 (P = 0.0002). The effects observed in MUA were also present in isolated single neurons (SOM). We found no correlation between the CI and the visual (P = 0.4385) or motor (P = 0.1204) properties of the neurons, indicating that both visual- and movement-related neurons (12) could be operantly controlled (Fig. 2E). The FR effect was also accompanied by a difference in the power spectra of local field potentials (LFPs) at the recording site. LFP power in the beta (13–30 Hz) and gamma (30–70 Hz) bands increased during Up versus Down trials (β median difference = 0.059 dB, P = 0.027; γ = 0.048 dB, P = 0.00068), while there was no difference for low-frequency theta (4–8 Hz) or alpha (8–13 Hz) bands (Fig. 2F; θ = 0.033 dB, P = 0.36; α = 0.063 dB, P = 0.33).
We wondered whether voluntary modulation of FEF activity might have associated effects on behavior, or on the function of the controlled neurons. Neuronal control could be achieved through nonspecific means such as changes in general arousal or vigilance, or perhaps through means that were motor in nature but that did not require a saccade (Figs. S4–S6). Alternatively, voluntary control might be achieved by a cognitive or behavioral strategy specific to the presumed role of FEF neurons (10, 14). We therefore employed visual search trials to probe psychophysical and neuronal performance during operant control. On 29% of operant control trials, the monkey was presented with a visual search array at an unpredictable time (Fig. 3A). Feedback then ceased, and the monkey was no longer rewarded for controlling neuronal activity. Instead, the monkey received a reward for directing a saccade to the search target (an oriented bar) if it was present at any location, or for maintaining fixation if no target was present.
Fig. 3
Fig. 3
Behavioral and physiological consequences of operant FEF control. (A) Visual search probe trials, in which a search array appeared and the monkey was rewarded (blue droplet) for directing a saccade toward an oriented bar target. (B) Percentage of probe (more ...)
In 82 experiments (Monkey B: 41, Monkey C: 41), the mean probe trial performance was 86.3% correct (Monkey B: 81.9%, Monkey C: 90.7%). The overall percentage of saccades targeting the RF was the same during Up and Down trials (Fig. 3B; Up = 22.6% of 4441, Down = 23.0% of 4346, P = 0.65; Monkey B: P = 0.43; Monkey C: P = 0.80). Likewise, the latencies of saccades to RF targets were similar during Up and Down trials (Fig. S7; z-score normalized Up latency = −0.001 ± 0.033; Down = 0.001 ± 0.032; P = 0.967). Thus, we found no evidence that operant control was associated with saccade preparation. The saccadic main sequence was also unaltered by voluntary control (Fig. S8).
In contrast, neuronal control had a spatially specific effect on visual search performance. The proportion of trials on which the monkey failed to detect the target in the RF (“misses”) was significantly greater during Down trials in both monkeys (Fig. 3C; Up = 5.7% of 1098, Down = 9.3% of 1124, P = 0.0017; Monkey B: P = 0.020; Monkey C: P = 0.032). In contrast, the proportion of misses at locations opposite the RF was statistically indistinguishable between Up and Down trials (Fig. 3D; Up = 2.7% of 1120, Down = 3.4% of 1114, P = 0.31; Monkey B: P = 0.39; Monkey C: P = 0.70). The influence of voluntary control on search performance was confined to locations less than ~6° from the controlled neuron’s RF (Fig. S9). We confirmed that search performance was specifically correlated with the direction of voluntary control and not with spontaneous fluctuations in pre-probe neural activity (Fig. S10 and SOM).
Next, we measured the effect of voluntary control on the ability of FEF neurons to identify the search target. As in previous studies (15) we found that the responses of FEF neurons (n = 150; Monkey B: 61, Monkey C: 89) could indicate whether a search target or a distracter appeared within the RF (Fig. 3E and SOM). However, we also found that the response to RF targets was 16.5% greater on Up than on Down trials (Fig. 3F; Up mean peak-normalized rate = 0.668, Down = 0.573, P = 0.0054). In contrast, there was no difference between responses to distracters during Up versus Down trials (Fig. 3G; Up = 0.265, Down = 0.313, P = 0.9207). Thus, neuronal control selectively enhanced FEF responses to targets but not to distracters, resulting in a significant enhancement in target discrimination during Up trials compared to Down (Fig. 3H; Up = 0.404, Down = 0.259, P = 0.004; Monkey B: P = 0.046; Monkey C: P = 0.037).
As with the behavior, target discrimination by FEF neurons was specifically related to operant control rather than non-controlled fluctuations in spontaneous pre-probe activity. Overall pre-probe activity for combined Up and Down trials did not predict the FEF responses to targets (median regression coefficient = 0.0145, P = 0.9337). However, dividing trials according to the direction of voluntary control revealed that pre-probe activity was positively correlated with the target response during Up trials (Fig. 4E,F; median regression coefficient = 0.0633, P = 0.0303), but anti-correlated during Down trials (median = −0.1128, P = 0.0468; Up versus Down: P = 0.0033). In contrast, responses to distracters were uncorrelated with pre-probe activity for both Up (Fig. 4G; median = −0.0072, P = 0.6560) and Down trials (median = 0.00, P = 0.9853; Up versus Down: P = 0.7431). Thus pre-probe FR combined constructively with target-driven activity only during upward control.
Fig. 4
Fig. 4
Correlation of spontaneous activity with FEF responses. (A) Linear regression between the probe trial response of an example FEF neuron and its spontaneous pre-probe activity during Up (red) and Down (blue) control. (B and C) Population histogram of regression (more ...)
Our results demonstrate that voluntary control of FEF neuronal activity is associated with spatially selective visual attention, measured behaviorally and neurophysiologically. In controlling FEF activity, monkeys converged on a strategy that dissociated spatial attention from saccadic preparation. This untrained dissociation provides a naturalistic demonstration that the two functions are not wholly interdependent, a point that has proven difficult to substantiate (10). We also observed that the attentional consequences of operant control were correlated with voluntarily driven, rather than spontaneous, FEF activity. This selective linkage might occur if control alters interactions between the FEF and visual cortex, a possibility supported by the enhanced LFP power at frequencies with suspected involvement in long-range interactions during attention (16). Finally, our results suggest a basis for recent evidence that neurofeedback training may be therapeutic in patients with attention-deficit hyperactivity disorder (17, 18) as they demonstrate that voluntary modulation of neural activity can indeed produce specific changes in cognitive function.
Supplementary Material
supplemental
Acknowledgments
We thank D. Aldrich for technical assistance, E.I. Knudsen, K.V. Shenoy for comments on the manuscript. This work was supported by NIH Grant EY14924, an NDSEG fellowship and National Research Service Award F31MH078490 (R.J.S.).
1. Cerf M, et al. Nature. 2010;467:1104. [PMC free article] [PubMed]
2. Fetz EE. Science. 1969 Feb 28;163:955. [PubMed]
3. Fetz EE. J Physiol. 2007 Mar 15;579:571. [PubMed]
4. Kobayashi S, Schultz W, Sakagami M. J Neurophysiol. 2010 Jan 27;103:1843. [PMC free article] [PubMed]
5. Carmena JM, et al. PLoS Biol. 2003 Nov;1:E42. [PMC free article] [PubMed]
6. Chapin JK, Moxon KA, Markowitz RS, Nicolelis MA. Nat Neurosci. 1999 Jul;2:664. [PubMed]
7. Fetz EE, Finocchio DV. Exp Brain Res. 1975 Sep 29;23:217. [PubMed]
8. Taylor DM, Tillery SI, Schwartz AB. Science. 2002 Jun 7;296:1829. [PubMed]
9. Bruce CJ, Goldberg ME, Bushnell MC, Stanton GB. J Neurophysiol. 1985 Sep;54:714. [PubMed]
10. Moore T. Curr Opin Neurobiol. 2006 Apr;16:159. [PubMed]
11. Supporting online material, including materials and methods, supporting text, and figures, are available on Science Online.
12. Bruce CJ, Goldberg ME. J Neurophysiol. 1985 Mar;53:603. [PubMed]
13. Sommer MA, Wurtz RH. J Neurophysiol. 2000 Apr;83:1979. [PubMed]
14. Tehovnik EJ, Sommer MA, Chou IH, Slocum WM, Schiller PH. Brain Res Brain Res Rev. 2000 Apr;32:413. [PubMed]
15. Thompson KG, Biscoe KL, Sato TR. J Neurosci. 2005 Oct 12;25:9479. [PMC free article] [PubMed]
16. Gregoriou GG, Gotts SJ, Zhou H, Desimone R. Science. 2009 May 29;324:1207. [PMC free article] [PubMed]
17. Gevensleben H, et al. J Child Psychol Psychiatry. 2009 Jul;50:780. [PubMed]
18. Loo SK, Barkley RA. Appl Neuropsychol. 2005;12:64. [PubMed]