The loss of latent inhibition across compound conditioning

Six experiments used a magazine approach paradigm with rats to investigate latent inhibition (LI). Experiment 1 found that compound conditioning did not increase evidence for LI, in contrast to predictions from acquisition-deficit models that are based on a common error term (e.g., J. M. Pearce & G. Hall, 1980; A. R. Wagner, 1981). Instead, it appeared that preexposed and non-preexposed stimuli conditioned to the same asymptote following compound conditioning, as is the case when these stimuli are conditioned separately. This was confirmed in three further experiments that used probe trials to measure conditioning to each conditioned stimulus across the course of compound training. In these experiments, LI was observed during initial but not extended compound training, consistent with predictions derived from M. E. Bouton (1997) and the SLG model (N. A. Schmajuk, Y. Lam, & J. A. Gray, 1996). However, 2 further experiments did not support these models. Instead, these findings are most consistent with models that use separate error terms to compute the associative strength of conditioned stimuli conditioned in compound (S. E. Brandon, E. H. Vogel, & A. R. Wagner, 2003; M. E. Le Pelley, 2004; N. J. Mackintosh, 1975).