Reflection suggests that neither view may be adequate. In the first place, as we have seen, escape is closely linked with the very different response system of ‘freezing’. Furthermore the terminating conditions of escape are often different from those of the response systems mediating attachment; thus the mere presence of the individual in a special location, or proximity to a mate, may each prove a haven of safety. Not only is ‘freezing’ very different from the behaviour patterns of crying, clinging, and following, but to be present in a location, if not to be in the proximity of a mate, is very different from the conditions which terminate attachment behaviour. Thus it seems useful for some purposes to distinguish two sets of instinctual response systems. Nevertheless, the discussion serves to show how intricately linked, through the existence of common activating and terminating conditions, these different systems tend to be and how misleading it would be were we to make a sharp division of them into two separate groups. Indeed, the adoption of a theory of instinctual behaviour such as that advocated here enables us to get away from any notion that each ‘instinct’ is entirely distinct from every other. Instead, it provides a flexible conceptual tool which promises to do justice to the complexities of the data.
So far we have been dealing only with those subjective experiences which accompany behaviour that is still at a primitive level. As conceived here, both primary anxiety and fright are the subjective components of instinctual response systems which are activated by certain conditions (part internal and part external, part unlearned and part learned by processes of conditioning) and which operate automatically. Not until the individual can structure his universe in terms of objects existing in time and space and causally related to one another can he develop the notion of a situation which is potentially dangerous. This leads us to differentiate a new class of behaviour with its own characteristic subjective accompaniment: these I shall term respectively avoidance behaviour and expectant anxiety.
As soon as the individual, whether human infant or a member of an infra‐human species, has reached a stage of development in which some degree of foresight is possible, he is able to predict situations as dangerous and to take measures to avoid them. In this he is exercising a far more complex function that is required for instinctual responses and one which Freud habitually attributed to the ego.
At least three sorts of danger situation are distinguishable, though for reasons already given there is some overlap between them. They are:
1 Situations in which the individual believe he is likely to be assailed by external stimuli which he finds (either ‘naturally’ or through learning or both) to be disagreeable and/or noxious and which, if realized, would activate his instinctual response systems of escape and freezing.
2 Situations in which the individual believes he is likely to lose that external condition which terminates his escape responses, namely his haven of safety.
3 Situations in which the individual believes certain of his instinctual responses will be activated without conditions for terminating them being likely to be present. Some such situations are already covered under (a) or (b); an example of one which is not is the prospect of sexual arousal in the absence of conditions for satisfaction.
The anticipation of any of these kinds of situation, and particularly the first two which appear to be the main ones, at once motivates him to take action intended to avoid their developing. Such ‘action’ may be of many kinds and will vary both in regard to the decisiveness with which a plan is made and in regard to whether or not it is actually executed. Irrespective of the mode of action resulting and irrespective, too, of which kind of danger situation is anticipated, the subjective states accompanying anticipation and avoidance appear to be the same: they are those of expectant anxiety.
The division of danger situations into two main classes, namely (a) and (b) above, is consistent with the empirical findings presented in a recent paper by Dixon, de Monchaux and Sandler (1957): a statistical analysis of patients’ fears showed that they tend to cluster into ‘fear of hurt’ and ‘fear of separation’.15 As these authors point out, moreover, it is consistent with Freud’s distinction between anxieties relating to castration and those associated with loss of object. It will be clear, however, that the two classes I have defined are more inclusive than Freud’s: in the scheme presented here castration anxiety and separation anxiety each represent a particular albeit important example of a broader class. The third class defined above, (c), was the first to be discussed by Freud and is present in his theorizing from 1894 onwards.
It may perhaps be asked why the term ‘anxiety’ has been chosen to denote, in combination with a qualifying word, two such different emotional states as are referred to by ‘primary anxiety’ and ‘expectant anxiety’. There are two reasons. First, as Freud pointed out (1926, p. 165), anxiety carries with it a note of uncertainty. This is true both of primary anxiety, where it is uncertain whether or not the individual will reach a terminating situation, and of expectant anxiety, where the subject is uncertain whether or not he can prevent the danger situation materializing. The second reason is that I believe both classes play a large part in the genesis of neurotic anxiety. A note on questions of terminology, with particular reference to Freud’s usage, will be found in the Appendix.
This is a convenient moment to attempt a summary. We have now differentiated three classes of situation and three classes of behaviour, together with the corresponding subjective accompaniments to which they commonly give rise. The word ‘commonly’ is of importance, since situations can evoke behaviour (and its corresponding subjective experience) only when the organism is in an appropriate state. In the following tabulation the organism is assumed to be in such a state:
| Situations | Behaviour | Subjective accompaniment |
| 1. Which activate an instinctual response system without providing for its termination | Persistent activation of response | Primary anxiety |
| 2. Which activate instinctual response systems mediating escape or ‘freezing’ | Escape or ’freezing’ | Fright |
| 3. Which, if no action is taken, it is anticipated will so develop that | ||
| instinctual response systems mediating escape or ‘freezing’ will be activated | ||
| the haven of safety will be lostan instinctual response system will be activated in conditions unlikely to provide for its termination | Avoidance | Expectant anxiety |
In real life more than one situation may be present at once and behaviour of more than one kind and level result. Thus at the sound of an air‐raid warning each member of a family may experience expectant anxiety in regard to the possibility of harm coming both to themselves and their loved objects and may take precautions accordingly; whilst the whistle of a bomb may excite both escape and clinging responses simultaneously. Although in them the function of foresight, dependent on an appreciation of causal relationships, may be well developed, the example serves to