Stimulus-preceding negativity is modulated by action-outcome contingency Hiroaki Masaki a, Katuo Yamazaki a and Steven A.

Transcription

1 CE: Satish ED: Swarna Op: Ragu WNR: LWW_WNR_4997 Cognitive neuroscience and neuropsychology 1 Stimulus-preceding negativity is modulated by action-outcome contingency Hiroaki Masaki a, Katuo Yamazaki a and Steven A. Hackley b We investigated the relationship between action-outcome contingency and stimulus-preceding negativity (SPN), a motivationally sensitive event-related potential. Neuroimaging studies have shown that insular cortex (a known source of the SPN) is more activated prior to rewards that are contingent on prior correct action than rewards that are given gratuitously. We compared two gambling tasks, one in which the participant attempted to guess the profitable key-press option (choice) and one in which rewards were simply given at random (no-choice). The SPN that developed in anticipation of feedback was larger in the choice condition, especially at right anterolateral sites. These findings suggest that the SPN specifically reflects the expectation of response reinforcement, rather than anticipatory attention toward emotionally salient stimuli. NeuroReport 00: c 2010 Wolters Kluwer Health Lippincott Williams & Wilkins. NeuroReport 2010, 00: Keywords: affective/motivational, event-related potentials, reinforcement, reward prediction, stimulus-preceding negativity a Faculty of Sport Sciences, Waseda University, Japan, b Department of Psychological Sciences, University of Missouri-Columbia, USA Correspondence to Dr Hiroaki Masaki, PhD, Faculty of Sport Sciences, Waseda University, , Mikajima, Tokorozawa, Saitama 359, Japan Tel/fax: ; masaki@waseda.jp Received 27 November 2009 accepted 1 December 2009 Introduction The learning and execution of goal-directed behavior exhibit an exquisite sensitivity to action-outcome contingency. In the classical view [1] the organism forms an association between a response emitted in a particular setting and its rewarding or punishing consequences. Neurobiological research has shown that the development of such associations involves dopamine-mediated alterations in synaptic plasticity [2]. Mesencephalic dopamine cells fire upon receipt of unexpectedly large rewards and then, as the action-outcome association becomes well established, during the anticipation of rewards [3]. Congruent with this account, neuroimaging studies have shown that a variety of brain areas are more active during the expectation of a reward that is contingent upon correct previous responding than one that is given gratuitously [4 8]. Among these brain areas, there are two that each of the just-cited studies agree upon as exhibiting this pattern the dorsal striatum and the anterior insula/frontal operculum. The insula/operculum region is relevant to the present research because it is one of the main generators of the stimulus preceding negativity (SPN; [9,10]), which is a putative electrophysiological index of reward anticipation. The SPN is usually largest over right prefrontal cortex, crescendoing as the arrival of a motivationally relevant stimulus draws near. Such stimuli can include money, performance feedback, evocative photos or electric shocks [9,11,12]. Consistent with the assumption that the SPN reflects activity within cortical portions of the This study was carried out at Faculty of Sport Sciences, Waseda University. reward system, this surface negative component is greatly reduced in patients whose dopamine systems have been compromised by Parkinson s disease [13]. It stands to reason that, if insular/opercular activation is greater during anticipation of response-contingent than gratuitous rewards, and if this cortical region does indeed contribute to the SPN, then SPN amplitude should reflect action-outcome contingency. Testing this hypothesis against the alternative that SPN merely indexes the anticipation of affectively salient stimuli is important for identifying the functional significance of this component. Our experiment was modeled after a functional magnetic resonance imaging (fmri) study in which participants perception of response-outcome contingency was manipulated [5,6]. Perceptual and motor factors were roughly equated but in one condition ( choice ) the participants were led to believe that their responses determined the monetary gain or loss. In the other condition ( no-choice ), they correctly understood that there was no such relationship. In the choice condition, Tricomi et al. [5,6] found greater activation in the caudate and downstream structures that included the anterior insula/inferior frontal gyrus. For our similar study, we predicted a larger SPN on choice than no-choice trials. Methods Participants Twelve, right-handed, neurologically normal, young women (aged 20 23, mean age 22.0 years) were tested in this event-related potential experiment. They were c 2010 Wolters Kluwer Health Lippincott Williams & Wilkins DOI: /WNR.0b013e bc3

2 2 NeuroReport 2010, Vol 00 No 00 Fig. 1 Choice +? +??? No-choice! +!!! 500 ms 2500 ms Fixation Stimulus onset Response Feedback Schematic illustration of the methods. The feedback displays in these examples show monetary gains for both choice and no-choice trials. paid for their participation and gave their informed consent. This study was approved by the Waseda University Ethics Committee. Stimuli and procedure Each trial began with the presentation of two boxes positioned to the left and right of a fixation cross (Fig. 1). Five-hundred milliseconds later, the boxes filled with either a pair of question marks (choice condition) or exclamation marks (no-choice). Upon seeing the paired exclamation marks, participants simply pressed the thumb button on the response box. They were told that they had no control over the outcome on these trials. By contrast, on the choice trials they were instructed to select either the first or second-finger key when they saw the paired question marks. They were told that if they succeeded in guessing the correct button on these trials there would be a monetary gain (win). As soon as participants responded in the choice condition, the fixation cross was replaced by a blue, left or rightpointing arrowhead, indicating the chosen option. In the no-choice condition, a blue diamond replaced the fixation cross to simply indicate that the response had been registered. The feedback was delivered to the participant 2.5 s after their response, informing them of the monetary outcome. The picture of an intact, 50 Japanese-yen coin indicated gain, whereas a broken coin represented loss. In the nochoice condition, either a pair of intact or broken coins were displayed in the boxes. In the choice condition, by contrast, one box showed an intact and the other a broken coin. An intact coin on the chosen side indicated that f 50 had been won; a broken coin on that side indicated a loss of f50 (roughly equal to U.S. $0.50). Before the experimental blocks there was a training session consisting of two blocks of 64 trials. The task was similar except that the feedback was noninformative (viz, gray-colored disks corresponding in size to the coins) and no money was won or lost. The experimental session consisted of five blocks 64 trials. Half of the trials within each block were choice and half were no-choice, selected randomly. Win and loss trials occurred at approximately a 50 : 50 ratio. In addition to any net winnings, participants were paid f1500 for their service. Recordings and data analysis The electroencephalogram (EEG) was recorded from 128 sites using Ag/AgCl electrodes. Horizontal electrooculograms (EOGs) were recorded from positions lateral to the left and right outer canthi. Vertical EOGs were obtained with electrodes above and below the left eye. EEG and EOGs were recorded with a bandwidth of DC to 51 Hz ( 3 db /octave) using a Biosemi Active Two system (Biosemi, Inc., Amsterdam, Netherlands) and a sampling rate of 256 Hz. Processing of the EEG was performed with Brain Vision Analyzer (Brain Products, Munich, Germany). The EEG was re-referenced to digitally linked ear lobes and corrected for ocular movement artifacts offline [14]. The EEG epochs were time-locked to feedback onset. After segmentation and signal averaging the waveforms were further low-pass filtered with a 12-Hz cutoff (rolloff: 12 db/octave). Mean amplitude of the SPN was measured within a window extending 200 ms before the onset of feedback, at electrode sites F3, F4, C3, C4, P3, P4, T7 and T8. The baseline for these measurements extended from 2400 to 2200 ms with respect to feedback onset. Per condition amplitude means were then subjected to a 3-way analysis of variance with factors of condition (two levels), caudality (four), and hemisphere (two). Where post-hoc comparisons were required, the Newman-Keuls test was used.

3 SPN and action-outcome contingency Masaki et al. 3 Statistical analyses were based on voltage measurements but, for illustrative purposes, scalp topographical maps of current source density (CSD) are shown. Laplacian transformations of the grand averaged, choice-minusno-choice difference waves were calculated using Brain Vision Analyzer for the interval extending from 207 to 4 ms before feedback. These computations involved spherical spline interpolation (order of splines, four; maximal degree of Legendre polynomials, 10; default Lambda, 1e-five) [15]. Questionnaire At the end of the experiment we administered a brief, 7- point, Likert-scale questionnaire, modeled after that of Tricomi et al. [6]. Participants rated how much control they felt they had over whether they won or lost money during the choice and no-choice trials ( sense of control ) and their degree of confidence in discerning a pattern to the correct responses ( degree of certainty ). Results Behavioral data Mean response time was significantly longer for the choice (M ± SEM: 1091 ± 136 ms) than for the no-choice condition (669 ± 49 ms; t(11)=4.29, P < 0.01). The number of errors, in the sense of confusing the thumb and finger buttons, was very low (1.3 ± 0.5 for the choice vs. 1.0 ± 0.2 for the no-choice) and did not differ between conditions (t(11)=0.40, not significant). Results for the post-experimental questionnaires confirmed the effectiveness of our manipulation. Participants rated their sense of control as significantly higher in the choice (3.25 ± 0.3) than in the no-choice condition (1.75 ± 0.3; t(11)=3.00, P < 0.05). Their degree of certainty about the existence of a pattern to the winning answers was also significantly higher in the choice than in the no-choice condition (4.58 ± 0.3 and 3.00 ± 0.4, respectively; t(11)=3.27, P < 0.01). Stimulus preceding negativity Figure 2 shows the grand-averaged SPN waveforms (top panel) and the CSD distribution for the difference between the two conditions (bottom panel). The SPN started developing just after the key-press response and gradually increased until feedback onset in both conditions. However, the waveform was clearly larger for the choice than for no-choice condition. As the CSD map shows, the amplitude difference between conditions was greatest at right, anterolateral scalp sites, consistent with the predicted contribution of right, insular/opercular cortex. The 3-way analysis of variance confirmed that SPN was larger in the choice than in the no-choice condition [F(1,11)=12.76, P < 0.005], and over the right than left hemisphere [F(1,11)=17.82, P < 0.005]. In addition, smaller amplitudes over temporal regions were documented [F(3,33)=3.77, e=0.68, P < 0.05; post-hoc test, P < 0.05]. No interactions were found. Discussion We investigated the relationship between the SPN and action-outcome contingency in a gambling task with two contrasting conditions, choice and no-choice. Participants rated their sense of control over the outcome and their ability to predict the winning answer as greater in the choice condition. Reflecting this perceived contingency, the SPN was larger in the choice than no-choice condition especially at right anterolateral sites. We predicted this result based on four relevant fmri studies that had shown greater activation of the anterior insula/frontal operculum region preceding action-contingent rewards as compared to gratuitously given rewards [4 8]. Our prediction was also based on evidence that the right insula/operculum contributes to the SPN that precedes motivationally significant stimuli [9,10]. It had been argued at one time that the increased activation within dopamine target structures when rewards are contingent on action might merely be because of the greater salience of such rewards or to the nonspecific arousal associated with making a movement [8]. However, the fmri study by Tricomi et al. [5,6] equated motor requirements across conditions but still found greater activation when outcomes were perceived by the participants to have been contingent on their responses. Our electrophysiological study was modeled on that experiment and our similar positive results support those conclusions. It is not the mere presence of movement that matters. Rather, what is critical is the perception that rewards and punishment are contingent upon certain actions. Our study provides further evidence that, among the various neocortical regions that contribute to the SPN, the right anterior insula and overlapping inferior frontal gyrus are especially sensitive to the anticipation of affective/motivational stimuli. Of course, this is likely to comprise only a portion of the brain regions that are activated during the anticipation of a task-relevant stimulus. Other regions may play a more general role in perceptual attention (e.g. intraparietal sulcus and frontal eye fields, [16]), accounting for the modality and visuospatially-specific SPN topographies that are sometimes observed [17,18]. Not surprisingly, the anticipation of stimuli that are neither motivationally relevant, response-contingent, nor perceptually demanding such as trial-by-trial task instructions is associated with only small, non-lateralized SPNs [12]. If the SPN is to play a useful role in the study of reward processing and dopaminergic disorders, it will be important to develop methods for isolating the insular/opercular generator. Conclusion The increased SPN amplitude at fronto-lateral sites observed in the choice condition supported our hypothesis: This motivationally sensitive component reflects

4 4 NeuroReport 2010, Vol 00 No 00 Fig. 2 5 μv 500 ms F3 F4 C3 C4 P3 P4 T7 T8 Response Feedback Response Feedback 207 ms 4ms μv/m μv/m μv/m 2 Choice No-choice Grand-averaged waveforms of stimulus-preceding negativity. Thick lines represent the choice condition and thin lines represent the no-choice condition. The lower panel depicts current source density maps of the difference wave (choice minus no-choice). Note the right-hemisphere predominance over frontolateral sites. anticipation of rewards or punishments that are perceived to be contingent upon prior action rather than attention toward salient stimuli, per se. Acknowledgement The authors would like to thank Ms Rie Shinohara for her help in collecting the data. This study was supported by a Grant-in-Aid for Scientific Research (C) from the Japan Society for the Promotion of Science and Waseda University Grant for Special Research Projects (2008A-505) to H. Masaki. References 1 Skinner BF. The behavior of organisms: an experimental analysis. Oxford: Appleton-Century; Tsai H-C, Zhang F, Adamantidis A, Stuber GD, Bonci A, de Lecea L, Deisseroth K. Phasic firing in dopaminergic neurons is sufficient for behavioral conditioning. Science 2009; 32: Schultz W. Predictive reward signal of dopamine neurons. J Neurophysiol 1998; 80: Tanaka SC, Doya K, Okada G, Ueda K, Okamoto Y, Yamawaki S. Prediction of immediate and future rewards differentially recruits cortico-basal ganglia loops. Nature Neuroscience 2004; 7: Tricomi EM, Delgado MR, Fiez JA. Hemodynamic responses to anticipated and unanticipated reward relevant targets in an oddball task. Program No , 2002 Neuroscience Meeting Planner. Orlando, FL:

5 SPN and action-outcome contingency Masaki et al. 5 Society for Neuroscience, Online. Proceedings Soc Neurosci 2002; 679:4 6 Tricomi EM, Delgado MR, Fiez JA. Modulation of caudate activity by action contingency. Neuron 2004; 41: Tricomi EM, Fiez JA. Feedback signals in the caudate reflect goal achievement on a declarative memory task. Neuroimage 2008; 41: Zink CF, Pagnoni G, Martin-Skurski ME, Chappelow JC, Berns GS. Human striatal responses to monetary reward depend on saliency. Neuron 2004; 42: Kotani Y, Ohgami Y, Kuramoto Y, Tsukamoto T, Inoue Y, Aihara Y. The role of the right anterior insular cortex in the right hemisphere preponderance of stimuluspreceding negativity (SPN): an fmri study. Neurosci Lett 2009; 450: Brunia CH, De Jong BM, van den Berg-Lenssen MM, Paans AM. Visual feedback about time estimation is related to a right hemisphere activation measured by PET. Exp Brain Res 2000; 130: Masaki H, Takeuchi S, Gehring WJ, Takasawa N, Yamazaki K. Affectivemotivational influences on feedback-related ERPs in a gambling task. Brain Res 2006; 1105: Van Boxtel GJ, Böcker KB. Cortical Measures of Anticipation. J Psychophysiol 2004; 18: Mattox ST, Valle-Inclan F, Hackley SA. Psychophysiological evidence for impaired reward anticipation in Parkinson s disease. Clin Neurophysiol 2006; 117: Gratton G, Coles MGH, Donchin E. A new method for off-line removal of ocular artifact. Electroceph Clin Neurophysiol 1983; 55: Perrin F, Pernier J, Bertrand O, Echallier JF. Spherical splines for scalp potential and current density mapping. Electroenceph Clin Neurophys 1989; 72: Corbetta M, Shulman GL. Control of goal-directed and stimulus-driven attention in the brain. Nature Rev: Neurosci 2002; 3: Ohgami Y, Kotani Y, Hiraku S, Aihara Y, Ishii M. Effects of reward and stimulus modality on stimulus-preceding negativity. Psychophys 2004; 41: Velzen JV, Eimer M. Early posterior ERP components do not reflect the control of attentional shifts toward expected peripheral events. Psychophys 2003; 40:

6 AUTHOR QUERY FORM LIPPINCOTT WILLIAMS AND WILKINS JOURNAL NAME: WNR ARTICLE NO: 4997 QUERIES AND / OR REMARKS QUERY NO. Details Required Author s Response No queries