The Neuropsychology of Anxiety

 Chapter 1

 Summary

In summary, we hold that the the septo-hippocampal system acts to detect conflict between concurrently available goals; that it resolves this conflict by increasing the weight given to affectively negative information by goal-processing areas; and that this increase in the weight of affectively negative information produces increases in anxiety in some tasks while reducing the effects of interference on memory in others.

Summary 2

We shall argue that the role of the septo-hippocampal system in the control of memory is to resolve conflicts between similarly, and highly, attractive concurrent goals and, hence, in many cases to reduce the effects of interference. Critically for the relationship of memory control to anxiety, we postulate that the septo-hippocampal system achieves this resolution by increasing the valence of affectively negative associations of those goals.

 Defensive Distance and Defensive Direction

The crucial point...is that the forms of behaviour that are appropriate when, for example, a rat must leave an area where there is a cat are quite different from those that are appropriate when a rat must enter an area where a cat has been or might be. ... Also important is Blanchard’s concept of defensive distance. Different forms of behaviour are appropriate at different distances from the cat (e.g. fight is only appropriate when the cat is close), but this ‘intensity’ variable is independent of the cat/cat odour distinction. The Blanchards included the cat-related behaviours in a ‘fear defence battery’ and the cat-odour-related behaviours in an ‘anxiety defence battery’. Having already distinguished ‘fear’ and ‘anxiety’ on purely theoretical grounds, they then showed that the anxiety but not the fear behaviours are affected by anxiolytic drugs. Consistent with this analysis, simple phobia in people is also insensitive to anxiolytics (Sartory et al. 1990).
The Blanchards themselves base this categorical distinction between fear and anxiety on whether the predator is actually or only potentially present. However, learning-theory analysis (Chapter 3) suggests, rather, that the critical issue is whether the behaviour functions to remove the animal from, or to facilitate entry into, a dangerous situation—i.e. whether active or passive avoidance is involved: in the former case, fear is involved, in the latter, anxiety. This categorical distinction of ‘defensive direction’ is separate from the smoothly graded dimension of ‘defensive distance’ mentioned above.
It should be noted, however, that defensive distance, as defined by the Blanchards, is a cognitive construct. It does depend on physical (or temporal) distance from a danger, but also on the perceived amount of threat: the greater the perceived threat, the greater is the physical distance from it that combines with perceived threat to produce a particular value of defensive distance.

 Anxiolytics Weaken Anticipatory Frustration

Second, a key feature of human anxiety lies in apprehension of the possibility of, not pain, but failure or loss of reward. There are a number of well-controlled experimental paradigms in which such ‘anticipatory frustration’ (Amsel 1962, 1992) can be measured in animals; and there is much experimental evidence that anxiolytics weaken this state (Gray 1977).

 Neuromodulator Mechanism

Putting together these new observations with the earlier ones, our theory now holds that anxiolytic drugs directly reduce the release of noradrenaline and serotonin into the septo-hippocampal system and directly alter the encoding of theta frequency via areas such as the supramammillary nucleus. These direct effects on modulatory systems then degrade (but do not totally eliminate) information processing in the septo-hippocampal system. In turn, the changes in septo-hippocampal function then account for the bulk of the behavioural effects of all clinically effective, centrally acting, anxiolytic drugs.

 Hippocampus as Context Mapping

In essence, the function these workers attributed to the hippocampus is that of forming, in subhuman animals, spatial maps. At the human level, it was proposed that this spatial mapping faculty becomes a more general faculty of cognitive mapping, applicable to sensory and conceptual domains of all kinds.
... Consistent with these findings from lesion experiments, we have shown recently that virtually all of several hundred cells recorded from the hippocampus proper and entorhinal cortex had task-related correlates in a non-spatial timing task (Young and McNaughton 1997). Most of these ‘timing fields’ were unique and virtually none was consistent with the coding of a ‘place field’ by the cell in question.
...we review extensive data which suggest that ‘place fields’ do not code unambiguously for places, change with non-spatial changes in the environment, and are not topographically mapped into the hippocampus. All of these data suggest that the ‘placeness’ of fields is not the result of uniquely spatial encoding. They also demonstrate that no structure receiving output from the hippocampus could, from that output alone, determine unambiguously and at all times where the animal is in space.
...The spatial environment and spatial location constitute the most important forms of context that mammals encounter.

 Behavioral Interrupt

More specifically...the septo-hippocampal system is seen as responding to stimuli associated with punishment and non-reward (i.e. to threat) by interrupting ongoing behaviour so as to allow information gathering (the ‘attention’ output of the behavioural inhibition system).

 Hippocampal Goal Inputs

Here we resolve these problems through a linked set of ideas: that all inputs to the hippocampus represent goals (conflating the stimuli, memories, responses, motor programs, and plans of the first edition); that the hippocampus detects when there is conflict between concurrently active goals (conflating the just checking, comparator, prediction, and mismatch functions of the first edition); and that, when conflict is detected, the hippocampus produces output which increases the valence of affectively negative stimuli and associations...
‘Goal’ as an explanatory concept in psychology has always been contentious (George and Johnson 1985). However, we use it simply in the sense described by Hinde (1982, p. 307; see also Towe and Luschei 1981):

Some authors have labeled behaviours as ‘directive’ or ‘goal directed’ on the sole criterion that variable means are used to achieve a consummatory situation . . . [but if] each type of behaviour is stereotyped, its cessation could be due merely to inhibitory effects consequent upon performances, rather than to error signals. . . . [In contrast] if rats are subjected to spinal or cerebellar operations so as to interfere with their motor coordination, they may nevertheless use quite novel movements to make errorless runs through a maze. The essential point here is that the new movements are not stereotyped, but selected from variable patterns in such a manner as to bring the animal nearer the goal. Furthermore, the new patterns are ‘directly and efficiently substituted without any random activity’. For our current purposes, the most important feature of the term ‘goal’ is that it can be distinguished from both ‘stimulus’ and ‘response’ in their pure sense—and represents a necessary conflation of the two. The goal of the rat running down the runway has both a stimulus (in this case, place) component and a response (the animal’s tendency to run towards it) or motivational component. Take either of these aspects away and it would cease to be a goal. As discussed in detail in Chapter 7, we see the firing of hippocampal cells as reflecting information about goals. Goals often have fixed locations in space, hence the place fields often found with hippocampal cells. But they may not, hence the non-spatial fields seen in the majority of cells in non-spatial tasks. Furthermore, they depend on the animal’s (as opposed to the experimenter’s) intent, hence the changes in the location of place fields when reinforcement contingencies change, and the fact that hippocampal cell firing does not always reflect the contingencies which experimenters think that they have embodied in the paradigm.

 Hippocampal Goal Comparison

...the hippocampus itself need only know that there are a number of goals, not precisely what they are...
When it receives information about only one prepotent goal, the hippocampal formation is, effectively, in ‘just checking mode’. When, in addition, it receives information about a second conflicting goal, it detects the multiplicity and effectively enters ‘control mode’. Note that no real change in the nature of processing occurs between these two modes. At all times information enters the comparator and is integrated, but when the summation of inputs passes some threshold it also produces output. The output from the hippocampal formation involves the return of information to those areas whose activity has given rise to conflict. At present we view the hippocampal output as providing a simple, quite general, affectively negative biasing function. The effect is to increase the valence of affectively negative stimuli and associations of stimuli (memories), and hence to shift approach–avoidance conflict in the avoidance direction both on the current and future occasions. The net result of this process is determined, however, not by the hippocampal formation but by its target structures. ... The septo-hippocampal system produces such an increase in negative weighting at every cycle of a series of computations which progressively increase bias (and hence suppress goals) until one or other goal is sufficiently predominant to take control of the motor mechanisms. This increase in negative bias has two consequences. It affects current motor output directly, and it affects future motor output indirectly through its biasing of associations (including those that are just being formed). Additionally, while its recursive operations continue to fail to resolve the conflict, the septo-hippocampal system produces additional output which engages exploratory mechanisms designed to resolve the conflict through the obtaining of new information about the environment (thus preventing the animal from getting locked into an impasse). We also presume that during consolidation, and particularly during sleep (Appendix 7), the hippocampus can direct a similar ‘internal exploration’, which in human beings may best be thought of as rumination. ... A final point to note is that goal conflict may result from the activation of quite different response systems (e.g. approach, avoidance) or from the activation of an essentially single response system (e.g. lever pressing) equally by two different stimulus sources (left lever, right lever).

 Hippocampal Forgetting

Given the power of recursive networks, what use might the hippocampal system make of them? Our general answer to this question is that the hippocampal system acts so as to increase the valence of adverse outcomes associated with items (locations in space, associative bonds in memory, or motor programs). This ‘association’ can be innate or acquired. In relation to memory, this general approach makes our position the antithesis of most others (including that of Eichenbaum et al. 1994, with whose views we other- wise have much in common) by supposing that the hippocampal formation acts, not to strengthen associations that are retained, but to weaken those that are discarded. Consistent with this position, there is evidence that the errors made in memory tests by patients with hippocampal damage over-represent items that would have been correct under slightly different circumstances. Such a pattern of errors cannot come from enhanced forgetting, as would be predicted if the role of the hippocampus were simply to facilitate the formation of (correct) memories. It is predicted, however, if this role is to eliminate competing (incorrect) memories. Similarly, our theory predicts that hippocampal deficits will be seen only in those tasks in which interference is actually overcome by control subjects, not in those in which interference is present but not overcome by controls.

 Panic as Undirected Escape

The periaqueductal grey is activated by an immediate predator, by an immediate dominant conspecific, by pain, and by high levels of carbon dioxide. As noted earlier, Graeff (1994) has suggested that activation of the periaqueductal grey is the basis of panic attacks. The flight (escape) elicited by stimulation of this region is undirected, so much so that the animal is likely to crash into obstacles rather than leap over them.
The dorsal periaqueductal grey is connected to the medial hypothalamus, where more sophisticated escape mechanisms appear to be located. This in turn is connected to the amygdala, which coordinates simple avoidance. In LeDoux’s formulation, the amygdala is then viewed as being connected to the hippocampus, which mediates complex, particularly inhibitory, avoidance. However, in Chapter 6 and Appendix 3 we find extensive reasons to see the anterior cingulate as representing several additional levels of the active defence system, and hence as lying directly above the amygdala in the ‘danger avoidance’ hierarchy. By contrast, the hippocampus proper and, above it, the posterior cingulate appear to be parts of a parallel ‘danger approach’ hierarchy.

 Comparison to LeDoux

Our analysis of the defensive system has, so far, closely followed those of LeDoux and Graeff (see Chapter 6 and Appendix 2). However, we differ from them in thinking that the hippocampus should not be viewed as, say, a source of contextual stimuli which can come to control the amygdala (in which case there should be strong hippocampal– amygdala connections rather than, as is predominantly the case, the other way round). Rather, we think of the hippocampus as a device which can resolve conflicts (as outlined above), not only between separate programs concurrently activated within the amygdala, but also between programs in the amygdala and those outside the defence system. It is important in this context, as with the case of mirror drawing, to emphasize that the septo-hippocampal system is involved in the resolution of goal rather than motor conflict.

 Mechanism for Hippocampal-Initiated Investigation

This can be achieved by way of the projection, described by Lavin and Grace (1994), from the nucleus accumbens, via the ventral pallidum and nucleus reticularis thalami, to virtually the entire thalamocortical sensory processing system. Analysis of this projection (Gray et al. 1997) suggests that activation of this subiculo-accumbens–nucleus reticularis pathway should provide an overall boost, by disinhibition, to sensory processing. Thus, two of the key outputs of the behavioural inhibition system (Fig. 1.1), interruption of ongoing motor programs and increased attention to the perceptual world, are perhaps achieved simultaneously by activating the subiculo-accumbens projection.

 References to Investigate:

• defensive distance: Blanchard and Blanchard: 1990a, Blanchard et al. 1993
• place fields: O'Keefe & Dostrovsky 1961, Nadel 1991, O'Keefe Nadel 1978