brief communications Single-neuron responses to ... - CiteSeerX

© 2001 Nature Publishing Group http://neurosci.nature.com


brief communications Single-neuron responses to emotional visual stimuli recorded in human ventral prefrontal cortex Hiroto Kawasaki1, Ralph Adolphs1,2, Olaf Kaufman2, Hanna Damasio1, Antonio R. Damasio1, Mark Granner1, Hans Bakken2, Tomokatsu Hori1 and Matthew A. Howard III1,2 1

Department of Neurology, Division of Cognitive Neuroscience, University Hospitals and Clinics, 200 Hawkins Drive, Iowa City, Iowa 52242, USA

2

Department of Surgery, Division of Neurosurgery, University of Iowa College of Medicine, 200 Hawkins Drive, Iowa City, Iowa 52242, USA Correspondence should be addressed to R.A. ([email protected])

Both lesion and functional imaging studies in humans1,2, as well as neurophysiological studies in nonhuman primates3, demonstrate the importance of the prefrontal cortex in representing the emotional value of sensory stimuli. Here we investigated single-neuron responses to emotional stimuli in an awake person with normal intellect. Recording from neurons within healthy tissue in ventral sites of the right prefrontal cortex, we found short-latency (120–160 ms) responses selective for aversive visual stimuli. The prefrontal cortex participates in linking perception of stimuli to the guidance of behavior4, including the flexible execution of strategies for obtaining rewards and avoiding punishments as an organism interacts with its environment 5 . Regions in the medial and ventral aspects of the frontal lobe seem especially important in relating information about external sensory stimuli to interoceptive information that represents emotional significance. For instance, studies of this region in monkeys show responses that are modulated on the basis of the rewarding value of the stimulus, independent of its perceptual properties3. Studies in humans using the lesion method1,5,6 and functional imaging2 show that ventral and medial prefrontal cortex guide decision-making and social behavior on the basis

Fig. 1. Single-neuron responses to emotional stimuli. (a) Mean (± s.e.m.) responses to emotional scenes recorded at medial or lateral locations in right ventral prefrontal cortex, as a function of emotion category. For each neuron, we calculated the absolute change in firing rate before (–2000 to 0 ms) and after (0 to +1000 ms) the stimulus, normalized with respect to the mean rate before the stimulus across all neurons. Thus, a percent change shown on the figure could reflect either a decrease or an increase in firing rate. Left, responses for all neurons (46 medially, 47 laterally). Right, responses for neurons that showed a significantly different rate in at least one pairwise comparision of emotion categories (thresholded at p < 0.05). Mean absolute spike rate change recorded medially differed significantly only when comparing aversive pictures to pleasant (p < 0.005) or to neutral b pictures (p < 0.005); at lateral sites, there were signifia cant differences only between mean rates for aversive and neutral pictures (p < 0.05; Wilcoxon rank-sum tests). (b) Responses to emotional facial expressions (37 neurons medially, 35 laterally), using the same analysis as in (a). Whereas mean absolute spike rate change between happy and fearful faces differed significantly medially (p < 0.05), it did not differ at lateral sites (p > 0.7). (c) Composite peristimulus-time histograms (top) and raster plots (bottom) from four individual neurons recorded medially in response to emotional scenes (stimulus onset, vertical gray line). Raster plots are shown for presentations of each of 11 aversive stimuli at the bottom. (d) Speed of categoc d rization reflected in response latency. Top, p-values (grayscale; Wilcoxon rank-sum test, uncorrected) showing that neuronal responses to aversive scenes differed from pre-stimulus baseline (–750 to 0 ms), for neurons recorded medially. x-axis, beginning of epochs after stimulus onset (10-ms steps); y-axis, width of epoch (10-ms steps). The sharp vertical ‘edge’ at 120 ms after the stimulus represents the time when the firing rate in time periods as small as 50 ms (that is, the window from 120–170 ms) differed at the p < 0.05 level from the rate during the pre-stimulus period. Bottom, p-values showing that neuronal responses to different emotion categories differed from one another, as a function of the width of a window beginning at 120 ms after the stimulus. Whereas aversive scenes differed from both pleasant scenes (top black line) and neutral scenes (middle gray line) at widths of 40 ms and 80 ms respectively, pleasant and neutral scenes did not differ from one another at any window (bottom light gray line).

nature neuroscience • volume 4 no 1 • january 2001

15



brief communications

of emotional value. However, the participation of this brain region in the processing of human emotion has not been investigated directly at the single-cell level. We recorded from 93 neurons at four sites on 2 depth electrodes (1 lateral, 1 medial), in the right ventral prefrontal cortex of a 48-year old man with a history of medically refractory epilepsy. The subject had a normal neuropsychological and psychiatric profile, and was oriented and attentive. He performed normally on tests of frontal lobe function, including the Iowa Gambling Task6, which has been directly linked to the operation of ventromedial prefrontal cortex1,6. With the subject’s informed consent, hybrid clinical-research depth electrodes7 were implanted stereotactically in the prefrontal cortex. The electrodes permitted isolation of neurons from high-impedance tetrode contacts using the DataWave system. (For detailed methods and neuroanatomy of the recording sites, see supplementary information.) Subsequent clinical evaluations using electroencephalographic (EEG) recording, as well as the outcome of the neurosurgical resection, confirmed a seizure focus in right motor/premotor cortex, distal to the sites from which we report recordings. We showed the subject pictures of visual scenes and facial expressions, selected from standardized databases (the International Affective Picture System and Pictures of Facial Affect, respectively). In right ventromedial prefrontal cortex, singleneuron responses to aversive visual scenes differed significantly from responses to pleasant or neutral scenes (Fig. 1a); likewise, responses to facial expressions of fear differed significantly from responses to facial expressions of happiness (Fig. 1b). These differential responses could not be explained by differences in the physical properties of our stimulus images, which were of equivalent luminance, size and color composition. Examination of raster plots and of the peristimulus-time histogram showed that aversive stimuli triggered a short-latency, transient inhibition followed by a prolonged excitation; no such responses were seen to neutral or to pleasant stimuli (Fig. 1c). Only for aversive scenes were there differences between response rates before and after the stimulus, and these differences became significant at 120 ms after stimulus onset, for time periods as short as 50 ms (that is, for the interval 120–170 ms; Fig. 1d, top). Furthermore, the response to aversive stimuli during this inhibitory period (from 120 ms after stimulus) differed significantly from responses to pleasant or to neutral stimuli during the same period, for windows as small as 40 or 80 ms, respectively (Fig. 1d, bottom). Thus, these data provide evidence that neuronal responses in human ventromedial frontal cortex are influenced selectively by the emotional content of visual stimuli within 120–160 ms, a form of rapid and coarse stimulus categorization. Whereas the speed of such emotion categorization is surprising, it is consistent with evoked potential studies in humans, which find visual responses in frontal cortex with latencies of 150 ms (ref. 8). Our findings are also consistent with functional imaging studies reporting responses to aversive visual stimuli in medial sectors of the prefrontal cortex9. Our data indicate categorization that seems to be specific to an ecologically salient category: stimuli related to fear, threat or dan-

16

ger. The responses we observed may be related to increased emotional arousal or increased allocation of attention to such stimuli; in all likelihood, arousal, attention and other cognitive processes would be affected together to prepare an organism for rapid behavioral response to stimuli that signal potential danger. Furthermore, our data showed visually different neuronal responses to aversive stimuli as early as 120 ms after the stimulus (Fig. 1c), but these differences were not statistically significant for windows smaller than 40 ms due to sparse firing rates (Fig. 1d). Only during neurosurgical intervention is there an opportunity to record single-neuron data from the human brain, and this constraint dictates the neuroanatomical locations from which recordings can be obtained. Whereas responses related to visual processing have been studied in a variety of brain structures in a neurosurgical setting10,11, we investigated emotion processing at the single-neuron level in human ventral prefrontal cortex. Our findings indicated that neurons in right ventromedial prefrontal cortex participate in encoding the emotional value of aversive visual stimuli. Surprisingly, we found responses with latencies as short as 120–170 ms to aversive stimuli, suggesting that prefrontal neurons can provide a rapid and coarse categorization of the emotion. Prefrontal responses may modulate visual information processing in other brain regions, for instance, by feedback to temporal cortices12, a function consistent with top-down influences on visual responses in that region13. In addition, we also found a longer-latency response to aversive stimuli, consisting of a prolonged increase in firing rate. Rapid and delayed response components may thus reflect an iterative function, whereby the prefrontal cortex guides the processing of emotional stimuli at multiple temporal scales. Possibly, such scales could encompass both covert processes as well as processing that contributes to conscious awareness of the emotion. Note: detailed methods information is available on the Nature Neuroscience web site (http://neurosci.nature.com/web_specials/).

ACKNOWLEDGEMENTS We thank I. Volkov, D. Tranel and N. Denburg for technical help. Supported by grants to R.A. from the Center for Consciousness Studies, the EJLB Foundation, and the Klingenstein Fund.

RECEIVED 14 JULY; ACCEPTED 21 NOVEMBER 2000 1. Damasio, A. R. Descartes’ Error: Emotion, Reason, and the Human Brain (Grosset/Putnam, New York, 1994). 2. Elliott, R., Dolan, R. J. & Frith, C. D. Cereb. Cortex 10, 308–317 (2000). 3. Rolls, E. T. The Brain and Emotion (Oxford Univ. Press, New York, 1999). 4. Fuster, J. M. The Prefrontal Cortex. Anatomy, Physiology, and Neuropsychology of the Frontal Lobe (Raven, New York, 1989). 5. Damasio, A. R. Phil. Trans. R. Soc. Lond. B Biol. Sci. 351, 1413–1420 (1996). 6. Bechara, A., Damasio, A. R., Damasio, H. & Anderson, S. W. Cognition 50, 7–15 (1994). 7. Howard, M. A. et al. J. Neurosurg. 84, 129–132 (1996). 8. Halgren, E. et al. J. Physiol. (Paris) 88, 51–80 (1994). 9. Northoff, G. et al. Cereb. Cortex 10, 93–107 (2000). 10. Kreiman, G., Koch, C. & Fried, I. Nat. Neurosci. 3, 946–953 (2000). 11. Ojemann, G. A., Ojemann, S. G. & Fried, I. Neuroscientist 4, 285–300 (1998). 12. Sugase, Y., Yamane, S., Ueno, S. & Kawano, K. Nature 400, 869–872 (1999). 13. Tomita, H., Ohbayashi, M., Nakahara, K., Hasegawa, I. & Miyashita, Y. Nature 401, 699–703 (1999).

nature neuroscience • volume 4 no 1 • january 2001

-1-

SINGLE-NEURON RESPONSES TO EMOTIONAL VISUAL STIMULI RECORDED IN HUMAN VENTRAL PREFRONTAL CORTEX Supplementary Information Hiroto Kawasaki, Ralph Adolphs, Olaf Kaufman, Hanna Damasio, Antonio R. Damasio, Mark Granner, Hans Bakken, Tomokatsu Hori and Matthew A. Howard III

M ETHODS Subject. All research was conducted under federally and institutionally approved protocols, with the subject’s informed consent. The subject had an entirely normal neuropsychological and psychiatric profile, as assessed with the WAIS-III, the Beck Depression Inventory, the MMPI-2, the Benton Facial Discrimination Task and other standard neuropsychological tests. He had normal intelligence, affect, personality, language and visuoperceptual function, and was fully oriented and attentive. The subject was taking analgesic and anti-seizure medication at the time of testing, but was fully alert (he was busy corresponding with his email when we were not testing).

Recording. Subject and pre-amps were enclosed in a grounded metallic net for shielding (Kuraray, Okayama, Japan). Bipolar differential recordings were obtained over the course of 2 days, beginning 4 days after the implantation of depth electrodes. Bandpass-filtered (2Hz–5KHz), pre-amplified signals from each of 8 neurophysiological channels, together with a trigger channel indicating the occurrence of stimuli, were recorded analog onto a multichannel tape for off-line analysis. Signals were subsequently filtered (300Hz–3kHz), thresholded at greater than four times the mean noise amplitude, digitized at 20 kHz, and sorted using DataWave’s Experimenter Workbench. All subsequent analyses were conducted using MATLAB.

Stimulus Presentation. Digital visual stimuli were presented in a darkened room on a 14-inch LCD computer screen (1 m directly in front of the subject) in randomized order (4-s duration; 5–8 s inter-stimulus interval) in 3 sessions of 20 stimuli each. The subject was quiet, alert and reclined comfortably in a hospital bed; gaze fixation and alertness were monitored continuously by one of the investigators watching the subject. We used two classes of visual stimuli: emotional scenes (for example, pictures of wartime), and emotional facial expressions (happy and fearful faces). The stimuli showing emotional scenes were selected from a set of standardized stimuli1 that had been extensively normed on a 10-point scale with respect to valence and arousal: pleasant stimuli (valence > 7; n = 16), aversive stimuli (valence < 2 and arousal > 6.5; n = 11), and neutral stimuli (7 > valence > 2 and arousal < 6.5; n = 33). Arousal and valence ratings by normal controls1 and by the subject himself are given in the table below (means ± s.d.) and were very similar. Quantitative

-2-

comparisons confirmed that stimuli from different categories did not differ significantly in terms of their average luminance or color composition.

Normative ratings:

aversive neutral pleasant

valence 1.5±0.2 4.3±1.1 8.0±0.2

Subject’s ratings: arousal 7.0±0.2 3.7±1.1 4.5±0.8

aversive neutral pleasant

valence 1.2±0.6 4.8±1.2 7.3±0.8

arousal 8.1±1.0 4.3±2 4.5±1.4

Face stimuli were selected from standardized pictures of prototypical facial expressions2 and consisted of 13 pictures of happy faces and 7 pictures of fearful faces. The subject recognized the emotion in all of these stimuli correctly.

S UPPLEMENTARY F IGURE

-3-

Legend for Supplementary Figure: Neuroanatomical location of recording sites shown in coronal MR scan. Four recording locations in the ventral prefrontal cortex are shown in red and were located at the tips of the electrodes. Each recording location indicated on the figure corresponds to a tetrode cluster of 4 adjacent contacts cut flush with the shaft’s surface. Electrodes consisted of Pt-Ir wires (Ø 40µm, impedance 50-200 kΩ ) within a teflon shaft (O.D. 1.25mm; Radionics, Inc., Burlington, MA)7 . Intracranial EEG recordings demonstrated that interictal epileptiform activity and seizure onsets localized to the right posterior dorsal frontal lobe, distal to the recording location. All recordings presented in this report were obtained from brain tissue that showed no anatomical or electrophysiological abnormalities as detected with high-resolution MRI and EEG.

R EFERENCES:

1. Lang, P. J., Oehman, A. & Vaitl, D. The International Affective Picture System (Univ. of Florida Press, Gainesville, Florida, 1988). 2. Ekman, P. & Friesen, W. Pictures of Facial Affect. (Consulting Psychologists, Palo Alto, California, 1976).