letters to the editor

anticipatory SCRs for good decks than for advantageous. (Penalties never cancel the of reward or punishment hidden in the bad decks (Fig. 1d–f). gain, as in decks C & D.) The immediate deck from which subjects are about to Results suggest that across both exper- tendency to prefer the high reward does select, depending on whether anticipato- iments, card selection is driven by long- not need to be opposed in order to ry SCRs reflect negative or positive somat- term consequences, whereas anticipatory achieve. Apparent and ultimate goodness ic states, higher anticipatory SCRs also SCRs are driven by the immediate act to coincide. There is no conflict. Normal sub- coincide with the long-term conse- be performed, independently of the posi- jects should prefer decks A & B. quences— of a long-term tive or negative long-term value of the In the original task, the higher antici- negative or positive outcome. When antic- decision. In the original gambling task patory SCRs preceded card turns from ipatory SCRs do not develop, a support experiments5,6, anticipatory SCRs were bad decks; by contrast, in the modified mechanism for making advantageous interpreted as correlates of somatic mark- task, higher anticipatory SCRs preceded decisions under conflict and uncertainty ers that bias individuals’ decision-making. turns from good decks. Because higher falls apart, as was critically demonstrated However, by changing the schedule of anticipatory SCRs related to decks carry- in patients with prefrontal damage6. punishments and rewards in Experiment ing the immediate higher magnitude of Another explanation for the finding 2, we observed an opposite pattern of reward or punishment, the authors argue would be that high-magnitude anticipato- SCRs. We conclude that SCRs in the stan- that “…anticipatory SCRs are driven by ry SCRs before good decks reflect a non- dard version of the gambling task do not the immediate act to be performed” and conscious danger signal related to the likely provide evidence for the role of somatic are irrelevant to the final outcome. We risk of a large penalty. Given the modified markers in decision making. agree that SCRs index a process driven by task’s design, such signals would not influ- the immediate act to be performed, but ence behavior because the conscious assess- Ian Tomb, Marc Hauser, Patricia this does not mean that anticipatory SCRs ment of the overall goodness of the decks Deldin and Alfonso Caramazza are unrelated to long-term decisions. would prevail. Either interpretation is in http://www.nature.com/natureneuroscience Department of , Harvard The authors believe that if somatic accord with the somatic marker hypothesis. University, Cambridge, Massachusetts markers drive the evaluation of the good- Somatic markers assist decision-making, 02138, USA ness or badness of the decks, then higher covertly or overtly, but are not engaged in e-mail: [email protected] anticipatory SCRs should precede picking every decision, and their ‘advice’ need not from bad decks (decks C & D) in the be heeded. We caution against the idea that modified task. They assume that somatic -based signals ‘decide’ for us, other REPLY—The authors report two studies. markers can only be negative and only than in extreme situations. In one, they used the standard version of precede options leading to a bad outcome. our gambling task and replicated our However, somatic markers can be either Hanna Damasio, Antoine Bechara and results. In the other, they used a new task, positive or negative, and under conflict Antonio R. Damasio with a superficial resemblance to the orig- and uncertainty they help reject or University of Iowa College of Medicine, inal, but with substantial conceptual dif- endorse an option of action. This suggests Department of , Division of ferences. They suggest that their results a possible interpretation of the authors’ Cognitive , 200 Hawkins Drive, are not compatible with our interpreta- results, namely that their task inverts the Iowa City, Iowa 52242-1053, USA tion of the original gambling task or with marker signal. In the original task, the e-mail: [email protected] © Group 2002 Nature Publishing the somatic marker hypothesis. We wel- higher anticipatory SCRs relative to bad come their question but disagree. decks reflected a negative somatic state 1. Damasio, A. The of What Happens The task in their second experiment that promoted avoidance of bad options. (Harcourt, New York, 2000). involves two decks of cards (A & B) for In the modified task, higher anticipatory 2. Panksepp, J. Affective Neuroscience (Oxford Univ. Press, New York, 1998). which both rewards and punishments are SCRs to good decks may reflect a positive 3. Rolls, E. T. Brain and Emotion (Oxford Univ. high, but in which the rewards always out- somatic state that promotes approach. Press, Oxford, 1999). 8,9 weigh the punishments; playing from these Our own work with another modified 4. Damasio, A. Descartes’ Error (Putnam, New decks offers both immediate and long- gambling task that preserved conflict and York, 1994). term gains. (A & B are ‘good decks’.) Decks uncertainty, but switched reward and 5. Bechara, A., Tranel, D., Damasio, H. & C & D have smaller rewards and punish- punishment, revealed approach behavior Damasio, A. Cereb. Cortex 6, 215–225 (1996). ments, but still the punishments are four coupled with high anticipatory SCRs. 6. Bechara, A., Damasio, H., Tranel, D. & times higher than the rewards; playing Also, patients with ventromedial pre- Damasio, A. R. Science 275, 1293–1294 (1997). from these decks gives low immediate frontal cortex lesions (which preclude the 7. Stout, J. C., Rodawalt, W. C. & Siemers, E. R. reward and long-term loss (‘bad decks’). development of these anticipatory SCRs) J. Int. Neurol. Soc. 7, 92–101 (2001). Compared to the original task, decks A & performed disadvantageously on that 8. Bechara, A., Damasio, H. & Damasio, A. R. B are more readily recognizable as prefer- modified task, just as they did on the orig- Cereb. Cortex 10, 295–307 (2000). able (immediate reward is 10 times the inal task. Although higher anticipatory 9. Bechara, A., Dolan, S. & Hindes, A. value of that in C & D) and ultimately SCRs probably relate to the magnitudes Neuropsychologia 40, 1690–1705 (2002).

1104 nature neuroscience • volume 5 no 11 • november 2002