QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

HOME  |   SEARCH  |   ARCHIVE  |   SUBSCRIBE  |   CONTACT  |   HELP

The Journal of Neuroscience, June 10, 2009, 29(23):7489-7496; doi:10.1523/JNEUROSCI.0349-09.2009

This Article Full Text Full Text (PDF) Supplemental Data Submit an eLetter Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Google Scholar Articles by Jocham, G. Articles by Ullsperger, M. PubMed PubMed Citation Articles by Jocham, G. Articles by Ullsperger, M.

 Previous Article  |  Next Article 

Behavioral/Systems/Cognitive
Adaptive Coding of Action Values in the Human Rostral Cingulate Zone

Gerhard Jocham,¹ Jane Neumann,² Tilmann A. Klein,¹ Claudia Danielmeier,¹ and Markus Ullsperger¹

¹Cognitive Neurology Research Group, Max Planck Institute for Neurological Research, D-50931 Cologne, Germany, and ²Department of Cognitive Neurology, Max Planck Institute for Human Cognitive and Brain Sciences, D-04103 Leipzig, Germany

Correspondence should be addressed to Gerhard Jocham, Cognitive Neurology Research Group, Max Planck Institute for Neurological Research, Gleueler Strasse 50, D-50931 Cologne, Germany. Email: jocham{at}nf.mpg.de

Correctly selecting appropriate actions in an uncertain environment requires gathering experience about the available actions by sampling them over several trials. Recent findings suggest that the human rostral cingulate zone (RCZ) is important for the integration of extended action–outcome associations across multiple trials and in coding the subjective value of each action. During functional magnetic resonance imaging, healthy volunteers performed two versions of a probabilistic reversal learning task with high (HP) or low (LP) reward probabilities that required them to integrate action–outcome relations over lower or higher numbers of trials, respectively. In the HP session, subjects needed fewer trials to adjust their behavior in response to a reversal of response–reward contingencies. Similarly, the learning rate derived from a reinforcement learning model was higher in the HP condition. This was accompanied by a stronger response of the RCZ to negative feedback upon reversals in the HP condition. Furthermore, RCZ activity related to negative reward prediction errors varied as a function of the learning rate, which determines to what extent the prediction error is used to update action values. These data show that RCZ responses vary as a function of the information content provided by the environment. The more likely a negative event indicates the need for behavioral adaptations, the more prominent is the response of the RCZ. Thus, both the window of trials over which reinforcement information is integrated and adjustment of action values in the RCZ covary with the stochastics of the environment.

---

Received Jan. 22, 2009; revised April 14, 2009; accepted May 8, 2009.

Correspondence should be addressed to Gerhard Jocham, Cognitive Neurology Research Group, Max Planck Institute for Neurological Research, Gleueler Strasse 50, D-50931 Cologne, Germany. Email: jocham{at}nf.mpg.de

This article has been cited by other articles:

				E. R. A. de Bruijn, F. P. de Lange, D. Y. von Cramon, and M. Ullsperger When Errors Are Rewarding J. Neurosci., September 30, 2009; 29(39): 12183 - 12186. [Abstract] [Full Text] [PDF]

Home  |   Search  |   Archive  |   Subscribe  |   Contact  |   Help