| |
Abstract:
Abstract: Stronger optimal (conscious) than sub-optimal (less
conscious) affective priming results (Murphy and Zajonc, 1993) were
simulated with a neural network model which forms a generalization
of the neurobiological 'dual-pathway' model for fear-conditioning
(LeDoux, 1995). Both 'quick and dirty' processing in the direct
pathway and 'slow and extensive' processing in the indirect pathway
can lead to fear reactions. A similar architecture is postulated
for the processing of positively and negatively valenced primes
(i.e. happy and angry facial expressions). Suboptimally presented
affective stimuli are primarily processed in the direct pathway,
whereas optimally presented stimuli are also processed in the
indirect pathway (Morris, Öhman & Dolan, 1999). In
suboptimal conditions, and in all conditions when the model was
instructed to evaluate the affective stimuli themselves, the model
produced congruent ratings. The incongruent priming effect found in
the optimal conditions by Murphy and Zajonc, however, resulted from
the more extensive processing in the indirect pathway. In a second
set of simulations, learning was introduced only in the indirect
pathway. Positive and negative stimuli were presented in an
emotionally relevant or irrelevant context. The Murphy and Zajonc
pattern of results emerged from this learning. The model not only
specifies conditions in which these results can or cannot be found,
but also strengthens the assumption of LeDoux (1995) that
regulatory control of emotional behavior can be acquired in the
indirect pathway.
|
