A model is presented that postulates that humour is the result of an unstable boost in total activation when two ordinarily competing interpretations are unified by a neural network. It is shown that this model is capable of explaining the four types of humour that result when the humour corpus is divided along two binary dimensions. The first dimension, that of incongruity/incongruity-resolution, splits humour into that which works by the mere juxtaposition of oppositions, and that which relies on an inference process to justify the presence of the incongruity. The second dimension concerns itself with whether the humour is innocent or tendentious (sexual, scatalogical or aggressive). The ubiquity of tendentious humour is accounted for by augmenting the model with units representing excitatory drive, and a competing force representing an inhibitory censor. The habitual action of the censor creates a high drive state, which provides non-specific arousal to the units subserving the competing interpretations when the censor itself is suspended by the humorous situation, thereby resulting in an augmented activation boost. Additional predictions of the model are also presented, showing why perceived humour is affected by the degree of expectation, resolution strength, timing and repetition. The paper concludes with a discussion of the relation of humour to pleasure in the context of the proposed model.
This paper is not available online