Elsevier

Neuropsychologia

Volume 37, Issue 4, 1 April 1999, Pages 403-411
Neuropsychologia

Ventromedial prefrontal cortex mediates guessing

https://doi.org/10.1016/S0028-3932(98)00107-9Get rights and content

Abstract

Guessing is an important component of everyday cognition. The present study examined the neural substrates of guessing using a simple card-playing task in conjunction with functional magnetic resonance imaging (fMRI). Subjects were scanned under four conditions. In two, they were shown images of the back of a playing card and had to guess either the colour or the suit of the card. In the other two they were shown the face of a card and had to report either the colour or the suit. Guessing compared to reporting was associated with significant activations in lateral prefrontal cortex (right more than left), right orbitofrontal cortex, anterior cingulate, bilateral inferior parietal cortex and right thalamus. Increasing the guessing demands by manipulating the number of alternative outcomes was associated with activation of the left lateral and medial orbitofrontal cortex. These data suggest that while simple two choice guessing depends on an extensive neural system including regions of the right lateral prefrontal cortex, activation of orbitofrontal cortex increases as the probabilistic contingencies become more complex. Guessing thus involves not only systems implicated in working memory processes but also depends upon orbitofrontal cortex. This region is not typically activated in working memory tasks and its activation may reflect additional requirements of dealing with uncertainty.

Introduction

The process of guessing involves making choice responses under incompletely specified situations. Guessing is an important component of everyday cognition and many laboratory cognitive tasks also involve some degree of uncertainty. In the most extreme case, the relationship between a response and desired outcome is entirely determined by chance. Real life examples of this are gambling situations; tossing coins, rolling dice, drawing playing cards from unbiased packs. In this study we used a gambling situation to determine the neural correlates of guessing.

Gambling tasks have recently been used in a series of studies by Damasio and colleagues 4, 5, 6, 7, 10. These studies found that patients with frontal lobe lesions, encompassing ventromedial regions, show pronounced impairments on gambling tasks. Strikingly, patients with lesions restricted to ventromedial foci show these impairments in the context of intact performance on most other cognitive tasks, including tests of working memory and planning [32]. The task used in these studies involved choosing one of four decks of cards, where one deck was relatively likely to be associated with positive outcomes and another with negative outcomes. Patients with ventromedial prefrontal lesions were unable to choose advantageously on this task despite demonstrating correct knowledge of the relevant contingencies 5, 6. They also failed to show the anticipatory autonomic response seen in normal subjects prior to receiving feedback for high-risk responses [5]. These findings suggest that although the orbitofrontal cortex is not essential for learning about contingencies per se, it is crucial for selecting the appropriate action based upon knowledge of these contingencies.

Despite neuropsychological findings suggesting a key role for the ventromedial frontal cortex in gambling or guessing tasks, there is relatively little functional imaging work addressing this issue. Two-choice tasks where there is no correct response have been used in the context of imaging studies of self-initiated or willed action 11, 18, 20, 21. In these studies, the selection of a particular response compared to the execution of a specified response was typically associated with activation of the dorsolateral prefrontal cortex but not the orbitofrontal cortex. However, the paradigms used in these studies differed critically from gambling tasks in that there was no feedback; that is the selected response was not subsequently deemed correct or incorrect. It therefore seems plausible that the role of the orbitofrontal cortex in guessing or gambling is not the selection of the response per se but the evaluation of the emotive consequences or behavioural significance of the selection. Consistent with this, our recent functional imaging study of the neural correlates of performance feedback [14]found differential activation of the ventromedial orbitofrontal cortex during performance of a guessing task in conditions where feedback was given compared to no feedback conditions.

In the present study, we aimed to use functional imaging to assess the neural correlates of a guessing task, in order to extend the neuropsychological findings. The task we used involved a simple gambling situation where subjects simply had to predict either the colour or the suit of the next card in an unbiased pack. In this situation, unlike the gambling task used by Damasio and colleagues, subjects cannot learn contingencies to allow them to make advantageous choices. Damasio and colleagues interpret their results in terms of decision making and working memory, an interpretation we agree with. However, other interpretations can be placed on Damasios findings; for example, the task can be considered in terms of associative learning, relating decks to outcomes. In our task, the correct response is entirely determined by chance on each trial and therefore there is no possibility of associative learning. However, the task still involves an element of risk and subjects were given feedback as to whether their prediction was correct on each trial. Our hypothesis was that orbitofrontal cortex, and particularly medial regions, would be activated in association with guessing.

Section snippets

Subjects

Five right-handed male volunteers, three male and two female, aged between 29–41 were recruited. Subjects who reported any neurological or psychiatric history were excluded, as were subjects with any evidence of abnormality on a structural MRI scan. The study was approved by the local hospital ethics committee and informed written consent was obtained prior to the study.

Cognitive activation paradigm

This experiment was a 2×2 factorial design with predicting compared to reporting as one factor and level of difficulty (colour

Performance data

In the guessing task, all five subjects adopted a probability matching approach, that is, in the guess colour condition they pressed the left and right hand keys with approximately equal frequency. In the guess suit condition, they pressed the left hand key (corresponding to their chosen suit) on approximately 25% of trials and the right hand key (corresponding to all other suits) on approximately 75% of trials. This suggests that subjects were complying with the instruction to guess which of

Discussion

The results of this study demonstrated that simple guessing was associated with significant changes in evoked activity in a network of structures, including the right dorsolateral prefrontal cortex, anterior cingulate, lateral orbitofrontal cortex, posterior parietal cortex, thalamus and cerebellum. There was also activation in the medial orbitofrontal cortex, confirming our hypothesis that this area is involved in gambling tasks. Consistent with this conclusion is the observation that this

References (32)

  • Baddeley AD. Working memory. Oxford: Clarendon Press,...
  • Baker SC, Rogers RD, Owen AM, Frith CD, Dolan RJ, Frackowiak RSJ, Robbins TW. Neural systems engaged by planning: a PET...
  • Barch DM, Braver TS, Nystrom LE, Forman SD, Noll DC, Cohen JD. Dissociation of working memory from task difficulty in...
  • Bechara A, Damasio AR, Damasio H, Anderson SW. Insensitivity to future consequences following damage to human...
  • Bechara A, Tranel D, Damasio H, Damasio AR. Failure to respond autonomically to anticipated future outcomes following...
  • Bechara A, Damasio H, Tranel D, Damasio AR. Deciding advantageously before knowing the advantageous strategy. Science...
  • Bechara A, Damasio H, Tranel D, Anderson SW. Dissociation of working memory from decision making within the human...
  • Berman KF, Ostrem JI, Randolph C, Gold J, Goldberg TE, Coppola R, Carson RE, Herscovitch P, Weinberger DR....
  • Courtney SM, Ungerleider LG, Keil K, Haxby JV. Transient and sustained activity in a distributed neural system for...
  • Damasio AR. Descartes error. New York: Putnam,...
  • Deiber M-P, Passingham RE, Colebatch JG, Friston KJ, Nixon PD, Frackowiak RST. Cortical areas and the selection of...
  • Dias R, Robbins TW, Roberts AC. Dissociation in prefrontal cortex of affective and attentional shifts. Nature...
  • Duvernoy HM. The human brain: surface, three-dimensional sectional anatomy and MRI. New York: Springer-Verlag,...
  • Elliot R, Frith CD, Dolan RJ. Differential neural response to positive and negative feedback in planning and guessing...
  • Fiez JA, Raife EA, Balota DA, Schwartz JP, Raichle ME, Petersen SE. A positron emission tomography study of the...
  • Fletcher PC, Josephs O, Friston KJ, Buechel C, Dolan RJ. Learning-related temporal dynamics in the prefrontal cortex....
  • Cited by (182)

    • The uncertain brain: A co-ordinate based meta-analysis of the neural signatures supporting uncertainty during different contexts

      2019, Neuroscience and Biobehavioral Reviews
      Citation Excerpt :

      Studies in this category typically require the participant to perform a forced decision with limited information with respect to its outcome. Various tasks in the literature are used to investigate decision-making under uncertainty, such as forced choice tasks for the most probable outcome, with a limited number of observed trials to learn, or with limited knowledge of, underlying probabilities (e.g., Krug et al., 2014; Volz et al., 2004), number or card prediction based on probability estimation (e.g., Elliott et al., 1999), category judgement based on limited observed trials or perceptual difficulty (e.g., Grinband et al., 2006; Seger et al., 2015), gambling tasks between low and more probable gain or high and less probable gain (e.g. Cohen et al. 2005), or reversal learning paradigms that operationalize uncertainty by requiring subjects to switch from a learned response to a different one when the contingent probabilities of the task unexpectedly change (e.g., D’Cruz et al., 2011; Robinson et al., 2010). Table 2 lists all the studies we included in this category.

    • Reasoning About Contingencies, Correlations, and Causes

      2018, Reasoning: The Neuroscience of how we Think
    • Reduced Neural Recruitment for Bayesian Adjustment of Inhibitory Control in Methamphetamine Dependence

      2016, Biological Psychiatry: Cognitive Neuroscience and Neuroimaging
    View all citing articles on Scopus
    View full text