EQ Institute Home Page

Includes:

1. Emotional capability, emotional intelligence, and radical change. Quy Nguyen Huy

2. Frontal lobe damage and decision - paper on Bechara, et al's gambling task etc

3. Abstract on some of Lanes work with Levels of emotional awareness


Mark Sainsbury

Frontal lobe damage and decision

1. Introduction

People with certain kinds of frontal lobe lesions (in particular ones in the ventromedial area) have difficulty with making decisions in real life (they adopt bad strategies, even when apparently knowing them to be bad, or they can't make up their minds), despite having good memory and general intelligence. They strike observers as having "low affect". However, they score well across a wide range of reasoning, personality and intelligence tests.

Damasio and colleagues successfully devised a laboratory test which would be distinctive of just this kind of damage: the Gambling Game. Participants were given (toy) money and asked to select a card from any one of four packs. They were told that they should try to make as much money as possible, and that each card turned would involve a pay-out and possibly also a fine. They were given no further information about the packs, and were not told how many selections they would be allowed to make (in fact they were allowed 100). Two of the packs, the "bad" packs, gave a high pay-out for each card turned ($100), but also imposed occasional high fines, so that every ten selections from that pack by a given player would net him a loss of $250. Two of the packs, the "good" packs, gave a lower pay-out for each card turned ($50), but the occasional fines were also lower, so that every ten selections from that pack by a given player would net him a gain of $250.

Damasio, Bechara and colleagues found that participants with ventromedial frontal lobe damage ("VM participants") tended to choose more cards (approx. 60%) from bad packs, whereas participants with other forms of brain damage and normals tended to choose more cards (approx. 63%) from good packs. The Gambling Game thus achieved its goal of providing an experimentally controllable functional mark of the relevant kind of damage.

In one version of the Gambling Game, the SCR (skin conductance response) of participants was measured both on receipt of a pay-out or fine, and in the period between the end of one turn and the selection of a fresh card (the "anticipatory SCR"). A surprising result was that although VM participants did not differ greatly from others in point of their SCRs in response to pay-out or fine, they had very markedly lower, indeed negligible, anticipatory SCRs. Moreover, the anticipatory SCRs of the successful participants tended to be higher before the selection of a bad pack than before the selection of a good one (nearly double).

The experimenters draw two conclusions:
In the Gambling Game, VM participants show an "insensitivity to future consequences" which replicates their real-life problems.
The data support the somatic marker hypothesis:
"The somatic marker hypothesis proposes that decision-making is a process that depends on emotion." (Bechara et al. 1999) In particular, it is suggested that the raised anticipatory SCRs in the case of successful participants acted as a warning to the subjects that they were about to make an unwise choice.

I argue that neither conclusion is well supported by the published data, which in fact support a different though no less interesting conclusion. This is that the Gambling Game poses an induction problem which is not easy to solve by explicit logical reasoning processes (for VM patients were good at such reasoning) but which can be solved by some other system, an affect-related one whose operation in successful participant is made manifest by their raised anticipatory SCRs. As far as the Gambling Game goes, there is no evidence that the VM participants had problems with the process of decision; a more modest hypothesis is simply that there is evidence that the VM participants did not manage to identify the relevant regularities.

2. Insensitivity to future consequences

Since all consequences are future ones, a natural way to understand the phrase "insensitivity to future consequences" involves a contrast: subjects are unduly sensitive to proximate rather than remote consequences. A paradigm would be the preference on the part of the majority of otherwise rational people for $100 today rather than $200 in a month.

No such contrast could influence the thinking of participants of the Gambling Game. For each selection, they want to maximize gain on that selection. If they have a pay-out and no fine, they will be pleased, the more pleased the bigger the pay-out. If they have a fine, they will be displeased, since that selection results in no gain and possibly in loss. There is no contrast between a proximate gain or loss and a remote (greater) gain or (lesser) loss.

Looking at the game from the experimenters' viewpoint, one can see how they might (wrongly) attribute this insensitivity towards the future. They may see a participant stuck into a strategy of selecting from a bad deck. They may know that this will deliver immediate gains which will be wiped out by future fines. But this is not how options could present themselves to a player with no explicit knowledge of the schedule of pay-outs and fines.

By the last quarter of the selections, most participants could "conceptualize" the pay-out/fine schedules, at least to the extent of recognizing the bad packs as "risky" and the good ones as "safe". It is striking that some of the VM participants also achieved this recognition, though it did not stop them from performing less well than normals. Though this fact calls out for explanation, unless the conceptualization were more detailed than "risky"/"safe", this would still not be properly described as insensitivity to future consequences: not all those who relish high risk are thus insensitive. In any case, only the minority of VM participants achieved this conceptual recognition, and even for them only a minority of their selections could have been guided by it (for the recognition came late in the game). So even if these minority selections did manifest insensitivity to future consequences, some other account would be needed of the majority of the results.

3. SCRs are not warnings

Normal subjects show a SCR as they are deliberating a response in the task. This anticipatory SCR appears to be a component of a warning to subjects that they are about to make a risky choice. As a result, they tend not to choose cards on which they could potentially lose money in the long term. [the VM participants suffer] a defect in using reward or punishment in the future to guide decision-making. The VM frontal subjects are thus defective at using prior experience to trigger somatic states and to guide choices." (Adolphs et al. 1996: 1623)

Higher SCR is followed by a selection from a bad pack. It is therefore not effective as a direct warning with respect to that selection. Relatively high SCRs cannot be thought of as inhibiting bad selections, since they are present shortly before bad selections and not shortly before good selections.

One can devise intelligible scenarios upon which the raised SCR acts as a warning ; but these are quite complicated, involving details not sustained by the evidence presented in the literature I cite. Such a scenario must take into account the fact that a selection from a "bad" back will more often than not be a resounding success (that is, it will result in high reward and no fine). If the memory of such an event includes both the fact that the selection was preceded by high SCR and the fact that it was successful, it is hard to explain why the success does not trump the other aspect. If the memory is of an unsuccessful selection from a bad pack, memory of the fine itself would seem to be the most effective deterrent; one would need an explanation of why memory of the fine needs supplementation by memory of raised SCR.

4. An alternative interpretation

The Gambling Game can be seen as an inductive problem: to be successful involves being guided by a hard-to-detect regularity. VM patients can certainly solve some inductive problems, as evidenced by their successful performance on the Wisconsin sort test. In these cases, the problem has a confined character: it does not require bringing to bear accumulated past data for a regularity, the proposed sort is either right or wrong, and the range of alternative sorting criteria is manifest. While it is not clear what the appropriate general feature of this kind of task is, let us for convenience say that the Wisconsin sort test can be solved by "explicit logical reasoning".

The VM patients cannot solve the induction problem posed by the Gambling Game. Hence this problem is not one which yields to explicit logical reasoning. So the hypothesis is that there is some other information-processing system, possessed by normals and patients with other kinds of brain damage, but lacked by VM patients, which will, in time, deliver an at least approximate solution to the problem. If there is such a system, there are two things we know about it: it can operate without engaging conscious awareness (for the performance of normal subjects began to improve even when they were unable to discover any conscious basis for distinctions among the packs); and there is a correlation between exercising it and at that time having raised SCR. The VM subjects, who on this hypothesis lack the relevant information-processing system, have flat SCRs at the time at which such a system ought to be operating; whereas successful subjects had raised SCRs during this period. Let us call the postulated alternative system for solving induction problems an "affect-related" one.

The proposed interpretation is that there are at least two systems for solving induction problems: explicit logical reasoning, which VM patients possess, and which will help with Wisconsin card sorting but not with the Gambling Game; and an affect-related system, which VM patients lack, but which will supply an approximate solution to the Gambling Game.

On this view, SCRs aren't signals to the subjects, but show that the subjects are using their affect-related inductive system.

The alternative interpretation is consistent with and explanatory of the data. The view that VM patients show insensitivity to the future is not explanatory of the data; nor is a simple-minded version of the view that raised SCRs act as warnings (and a suitably complex version has not, as far as I am aware, appeared in print). So the experiments do not support the view that raised SCRs are "somatic markers" in this sense. There is other evidence for the interpretation proposed here.

Two systems.

There are many cases of duplication of systems. More specifically, there are many cases in which what counts, within a useful taxonomy of functions, as a single function is subserved by distinct neurological mechanisms. A clear case is fear, as described by Le Doux (1998).

In the case of induction, there is common sense evidence of two systems. There is a big difference between a good "intuitive" bridge or poker player and one who explicitly keeps track of cards and odds. There is a big difference between the hunches of someone familiar with a subject matter and explicit and often laborious statistical reasoning. One measurable difference is that hunches generally deliver more quickly. VM participants took longer to choose their pack than did successful participants.

Wason selection task.

People are generally bad at saying what additional information they need to check in order to test whether a partially described state of affairs falsifies a general statement. However, normal subjects do better under special conditions containing affect-involving subject matters, variously described as related to "cheat detection", "familiarity" or "the social world". "Doing better" is being more accurate, and the responses were also quicker. This suggests that, in normals, affecting subject matter can engage better reasoning processes. Such reasoning processes are in some special way related to the affecting nature of the subject matter. It is striking that VM subjects do not perform the Wason selection task better under these special conditions (Adolphs et al 1996). This suggests they have a general deficit of affect-related information-processing mechanisms, thus supporting the present interpretation of their behaviour in the Gambling Game.

This interpretation is congenial to the general line that Damasio takes in Descartes Error. It suggests that emotion (affect-related information processing mechanisms) rather than reason (explicit logical reasoning) is needed even in tasks which might otherwise be supposed to lie wholly within the province of reason. It differs from Damasio only on the details: in this interpretation, there is no mention of insensitivity to future consequences and raised SCRs are not construed as warnings, or as themselves purveyors of information.

References:

Adolphs, R., D. Tranel, A. Bechara, et al. (1996). "Neuropsychological approaches to reasoning and decision-making." In A.R.Damasio, Y. Christen and H. Damasio (eds.) Neurobiology of Decision, 157179, New York: Springer.

Bechara, Antoine, Antonio R. Damasio, Hanna Damasio and S. Anderson (1994). "Insensitivity to future consequences following damage to human prefrontal cortex." Cognition 50, 712.

Bechara, Antoine, Hanna Damasio and Antonio Damasio (2000). "Emotion, decision making and the orbitofrontal cortex." Cerebral Cortex 10, 295307.

Bechara, Antoine, Hanna Damasio, Antonio R. Damasio and Gregory P. Lee (1999). "Differential contributions of the human amygdala and ventromedial prefrontal cortex to decision-making." Journal of Neuroscience 19.13, 54735481.

Bechara, Antoine, Hanna Damasio, Daniel Tranel and Antonio R. Damasio (1997). "Deciding advantageously before knowing the advantageous strategy." Science 275, 12931294.

Damasio, Antonio (1994). Descartes' Error. New York: Putnam.

LeDoux, Joseph (1998). The Emotional Brain. London: Weidenfeld and Nicolson.

Tranel, Daniel, Antoine Bechara and Antonio Damasio (1999). "Decision making and the somatic marker hypothesis." In Gazzaniga, M.S. (eds.) The New Cognitive Neurosciences, 10471061, Cambridge, MA: MIT Press.



Abstract - "Neuro-anatomical Correlates of Emotion" Richard D. LANE, M.D.

University of Arizona, Tucson, AZ and The Samaritan PET Center, Phoenix, AZ, USA

Objective:

Happiness, sadness and disgust are three emotions that differ in their valence (positive or negative) and associated action tendencies (approach or withdrawal). PET imaging during these emotions permits exploration of how different emotions are organized in the brain. In addition, we examined the neural correlates of the conscious experience of emotion by correlating activation patterns associated with emotion with scores on the Levels of Emotional Awareness Scale (LEAS), a measure of the complexity of conscious emotional experience.

Method:

Twelve healthy female subjects were studied. Positron emission tomography and 15O-water were used to measure regional cerebral blood flow (BF), a marker of local neuronal activity. There were twelve conditions per subject: happiness, sadness and disgust and three control conditions, each induced by film and recall. Emotion and control tasks were alternated throughout. Condition order was pseudo-randomized and counterbalanced across subjects. The LEAS was administered prior to PET imaging.

Results:

Discrete emotions were examined by combining emotions across stimulus conditions, e.g. film- and recall-induced happiness were averaged and contrasted with the average of film- and recalled-induced neutral states. A distributed set of highly significant and symmetrically organized areas of activation was observed. Thalamus and medial prefrontal cortex (BA 9) were activated by all three emotions. Anterior and posterior temporal structures were also activated by all three emotions, primarily attributable to film-induced emotions. Orbitofrontal cortex activity was greater in happiness than disgust or sadness and cerebellar vermis activity was greater in sadness and disgust compared to happiness. Activation of the anterior insula was greater during recalled sadness relative to happiness or disgust.

BF changes specifically attributable to emotion were then correlated with LEAS scores to identify regions of the brain that are associated with the conscious experience of emotion. Correlations between LEAS and BF change due to emotion (3 emotions minus 3 neutrals) were each significant (r>.63, p<.005) in anterior cingulate cortex for both the film-elicited and recall-elicited conditions. However, when correlating LEAS directly with the components of this difference, i.e. the 3 emotions alone or the 3 neutrals alone, correlations between BF and LEAS were not significant in the anterior cingulate for either the film conditions or the recall conditions (four alternative correlations).

Conclusions:

The present results demonstrate that happiness, sadness and disgust share common neuroanatomical substrates. The findings did not support current models of differences between emotions derived from previous neurological or electrocortical data. Discrete emotions share a common neuroanatomical basis, yet are also associated with significant differences in activation patterns.

The correlations with emotional awareness are consistent with recent findings indicating that BF in the anterior cingulate increases as a function of attention or conscious awareness to cognitive stimuli. These data also provide further support for the validity of the levels of emotional awareness construct.

Review of Lane chapter in Bar-On book