Standard buffer types do not distinguish between different types of data. However, many cognitive theories postulate specialised, domain-specific modules, with equally domain-specific properties.
This talk describes a specialised buffer for one-dimensional imagistic representations: the one-dimensional analogue buffer. This buffer type accepts only points and lines as input data, provides extra access properties, and implements a specialised form of imagistic decay, random point movement, after a suggestion by Logie (1995) that mental imagery cannot represent exact location.
This buffer has been used for an implementation of Berendt's (1996) model of reasoning in the domain of Allen Inferences. This model embodies a strategy to overcome the inexactness problem, and in so doing predicts certain response biases in human reasoners' solutions to these problems.