ECVP 2005 Abstract
doi:10.1068/v050495

Cite as:
Mogi K, Sekine T, Tamori Y, 2005, "Slow and fast processes in visual 'one-shot' learning" Perception 34 ECVP Abstract Supplement

Slow and fast processes in visual 'one-shot' learning

K Mogi, T Sekine, Y Tamori

A striking example of active vision is when it takes a while to realise what is hidden in a seemingly ambiguous bilevel quantised image. Famous examples such as 'the Dalmatian' and 'Dallenbach's cow' are visual teasers in which naive subjects find it difficult to see what is in the figure. Once the subjects realise what is in the picture, there is a one-shot perceptual learning, and recognition of what is in the image is possible after the passage of a considerable amount of time. These visual 'aha!' experiences or 'one-shot' learning processes are interesting for several reasons. First, the combination of low-level spatial integration and top - down processes involved provides important clues to the general neural mechanism of active vision (Kovaćs et al, 2004 Journal of Vision 4 35a). Second, the temporal factors involved in this process, such as the brief synchronisation of neural activities (Rodriguez et al, 1999 Nature 397 430 - 433), provide crucial contexts for the integration of sensory information. Here, we report a series of experiments where the temporal factors involved in one-shot visual learning are studied. The subjects were presented with several bilevel quantised images, and were asked to report what was in the image. We measured the time required in delivering the correct answer. We found that there are at least two distinct cognitive processes involved. In the 'fast' process, the subjects almost immediately realise what is in the image, with the report time distribution decaying in an exponential manner. In the 'slow' process, the realisation occurred in a quasi-Poisson process, with the moment of realisation evenly distributed over time. Thus, the visual system seems to employ at least two different strategies in deciphering an ambiguous bilevel quantised image. We discuss the implications of our result for the neural mechanisms of dynamic cognition in general.

These web-based abstracts are provided for ease of seaching and access, but certain aspects (such as as mathematics) may not appear in their optimum form. For the final published version of this abstract, please see
ECVP 2005 Abstract Supplement (complete) size: 2356 Kb