The most influential network theory of emotion was proposed by Gordon Bower (Bower, 1981, 1992; Bower & Cohen, 1982; Bower & Forgas, 2000). Based on the earlier Anderson and Bower (1973) Human Associative Memory (HAM) model, Bower proposed that concepts, events, and emotions can all be represented as nodes within a network. In fact, the type of network originally chosen by Anderson and Bower consisted only of labelled links; the nodes themselves had no semantic labels. However, for ease of explication the network illustrated in Figure 3.7 uses the system of labelled nodes rather than links; the two approaches are in fact computationally equivalent (see Johnson-Laird et al., 1984).
The network presented in Figure 3.7 shows that concepts, events, and emotions are all represented as nodes within the network. Activation within the network depends on a number of factors that include the proximity of nodes to each other, the strength of the initial activation, and the time lapse since activation. In the example shown, the Depression Emotion Node, or "DEMON" for short, possesses a wide variety of types of links, such as links to phenomenological and physiological characteristics, linguistic labels, and depression-related events and memories; Bower and Cohen (1982) suggest that some of the expressive behaviour and autonomic links may be innate.
Activation of one of the nodes in the network shown in Figure 3.7 may spread therefore to adjoining nodes; for example, if an individual experiences failure, the experience could activate the DEMON node and spread to a range of adjoining nodes, including activating previous events in which the individual experienced failure and activating innate and learned expressive and autonomic patterns. In a passage of purple prose that is more reminiscent of nineteenth-century science and Freud's early hydraulic model of psychic energy, Bower describes this activation as follows:
EVOKING -DEPRESSION _ (Inhibition) HAPPINESS
APPRAISALS EMOTION NODE EMOTION NODE
Figure 3.7 An example of a semantic network (based on Bower & Cohen, 1982).
The lines represent water pipes, the nodes represent reservoirs, and activation is like water that is pushed down the pipes with more or less pressure, with the water accumulating at the units where the lines come together. (Bower, 1986, p. 24)
Bower's network theory initially gave rise to an impressive collection of supportive evidence based primarily on the effects of the temporary induction of happy or sad moods in normal individuals. Bower (1981) reported data that showed that mood induction led to mood-state-dependent memory, and that it influenced a range of processes including free association, reported fantasies, social judgements about other people, and perceptual categorisation. For example, in a typical experiment Bower (1981) induced sad or happy moods in a group of hypnotisable subjects and then asked the subjects to recall childhood incidents. The results (see Figure 3.8) showed that when the subjects were in a happy mood, they recalled significantly more pleasant events from childhood, but when the subjects were in a sad mood, they recalled significantly more unpleasant events. On the assumption therefore that the subjects were in a matching mood at the time of the experience of the events (that is, a happy mood for a pleasant event and a sad mood for an unpleasant event), this study can be interpreted as evidence for mood-state-dependent memory.
Was this article helpful?