More than previous chapters, this one has been speculative. Based on the observation that behaviourally spontaneous confabulators adhere to their false ideas even when true reality contradicts them, I derived an analogy with deficient extinction, the capacity to learn from the non-occurrence of anticipated events. I invoked extinction rather than reversal learning for a reason: reversal learning occurs when there is an outcome, albeit inverse to expectation. In contrast, extinction occurs when an anticipated outcome fails to happen. In normal thinking, many ideas (anticipations) never...
More than previous chapters, this one has been speculative. Based on the observation that behaviourally spontaneous confabulators adhere to their false ideas even when true reality contradicts them, I derived an analogy with deficient extinction, the capacity to learn from the non-occurrence of anticipated events. I invoked extinction rather than reversal learning for a reason: reversal learning occurs when there is an outcome, albeit inverse to expectation. In contrast, extinction occurs when an anticipated outcome fails to happen. In normal thinking, many ideas (anticipations) never encounter an outcome. This may be the nature of fantasies: Fantasies differ from thoughts, which relate to ongoing reality, by the absence of real outcomes rather than unexpected outcomes. Behaviourally spontaneous confabulators cannot make this difference.
The idea that the reward system, by virtue of its extinction capacity, acts as a generic reality checking system requires that the concepts of reward and extinction be extended to unconscious anticipations and outcomes devoid of hedonic value. As regards the latter requirement, we have found strong activation in orbitofrontal areas 13 and 10 when healthy subjects anticipated and monitored simple outcomes that had no tangible reward value (Schnider et al. 2005b). The extension of the reward concept to unconscious processing does not appear to be more difficult. In decision-making, somatic reactions (somatic markers) can be detected even before a subject has conscious knowledge about which choice would be advantageous (Bechara et al. 1997). Why should the same principle not apply to anticipation and outcome processing in general? To apply the notions of reward processing and extinction to human reality checking simply requires one to acknowledge that human thinking and anticipation underlie the same physiological principles as reward processing by non-human primates, although human thinking is much more complex.
The ventromedial prefrontal area with the orbitofrontal cortex provides the neural apparatus and the connections allowing it to check the non-occurrence of events and to signal it to other parts of the brain. Monkeys with lesions of area 13 had a specific deficit of extinction, whereas more lateral lesions in the inferior convexity of the frontal lobes induced a deficit of object reversal learning (Butter 1969). Single cell recordings from area 13 demonstrated the presence of neurons firing specifically in extinction trials, when rewards failed to be delivered at the expected time (Rosenkilde et al. 1981). Conversely, neurons in the lateral inferior prefrontal cortex preferentially signalled the detection of reward.
Behaviourally spontaneous confabulation occurs not only after damage to area 13 but the whole posterior ventromedial prefrontal area. Within this region, an orbital and medial network have been distinguished (Carmichael and Price 1996). Both networks have similar, albeit separate, connections with structures whose damage may induce behaviourally spontaneous confabulation and which are involved in reward processing.
Both the medial and orbital networks have connections with dopaminergic structures in the ventral striatum and the mesencephalon. Dopaminergic neurons code the non-occurrence of anticipated rewards by phasic decrease of their firing rate (Schultz et al. 1997). Whether this decrease reflects primary processing of non-events by these neurons or their inhibition by a signal from the medial orbitofrontal cortex, has not been resolved.
Single cell recordings have shown that area 13 has a particularly high density of neurons signalling the absence of rewards, while lateral orbitofrontal cortex has a particularly high density of neurons signalling the occurrence of rewards. However there is no strict separation; both areas contain both types of neurons (Rosenkilde et al. 1981). If the hypothesis is correct that the capacity to adapt thought and behaviour to ongoing reality critically depends on extinction, and hence on the presence of extinction cells, then the findings from single cell recordings might explain some clinical mysteries. As we have seen in Chapter 4, only a small minority of patients, who have the diseases and lesions commonly responsible for behaviourally spontaneous confabulation, will actually go on to have the syndrome of behaviourally spontaneous confabulation. This is true for ruptured aneurysms, diencephalic tumours, traumatic brain injury, and alcoholic Korsakoff syndrome. Could it be that these patients had the misfortune of concentrating their extinction cells in the critical area around the posterior medial orbitofrontal cortex? And could it be that the availability of extinction cells in surviving tissue determined the speed of recovery from behaviourally spontaneous confabulation? Finally, could it be that the propensity of patients to act out their false beliefs reflected goal-seeking behaviour, whose intensity depended on the density of reward-signalling neurons in undamaged orbitofrontal cortex? Thus, could there be an anatomical predisposition for losing the ability to adapt thought and behaviour to ongoing reality following posterior medial orbitofrontal damage?
The hypothesis exposed here seeks to explain the reality confusion that some patients experience after orbitofrontal damage, not necessarily the tendency to talk about the falsely perceived reality, that is, the intensity of the confabulations. In contrast to the observations described in the previous chapter, which were derived from controlled evidence obtained from behaviourally spontaneous confabulators, the hypothesis explained in this chapter awaits controlled exploration in patients who confuse reality.
For future studies on any type of confabulation, it will be crucial to characterize precisely the patients and their clinical course, analyse the circumstances inciting the confabulations, investigate the range of confabulatory themes, describe associated behaviours, and devise specific experimental procedures to test the physiological basis of the suggested mechanisms.
Korsakoff's (1891) call for research on confabulations is as justified nowadays as it was almost 120 years ago: Research on the connections among latent [memory] traces and the influence of latent (unconscious) traces on the flow of ideas might help to explain many interesting phenomena concerning normal and pathological psychology. Korsakoff (1891, p. 410)
Chapter. 7743 words. Illustrated.
Full text: subscription required