Historically a role for the

Historically, a role for the prefrontal cortex in spatial working memory has been repeatedly claimed (for review, see Kolb, 1984). Older studies using traditional, non-fibre-sparing lesions (e.g. aspiration lesions) often resulted in spatial alternation deficits. In contrast, more recent studies employing cytotoxic, fibre-sparing techniques have found that lesions of medial prefrontal cortex (mPFC) do not always reliably affect non-matching to place performance in rodents, and where deficits do exist they are generally mild and transient in nature (Shaw and Aggleton, 1993; Aggleton et al., 1995; Sanchez-Santed et al., 1997; Delatour and Gisquet-Verrier, 2000; Dias and Aggleton, 2000; Deacon et al., 2003; Walton et al., 2003; Kellendonk et al., 2006). More recent studies have clearly shown that rodents with mPFC lesions are perfectly capable of learning the spatial non-match to place rule (e.g. Touzani et al., 2007), and can perform perfectly well even at longer delays (Gisquet-Verrier and Delatour, 2006). Indeed, Gisquet-Verrier and Delatour (2006, p. 585) have suggested that the mPFC “is not directly involved in the short-term maintenance of specific information but is implicated when changes, such as the sudden introduction of a delay or exposure to unexpected interfering events, alter the initial situation” (see also Touzani et al., 2007). This is in obvious TAME to the robust and lasting spatial working memory impairments that have been observed following lesions of the septo-hippocampal formation, or following GluR-A deletion (e.g. Deacon et al., 2002; Reisel et al., 2002).
To summarise, GluR-A−/− mice exhibit a profound and lasting impairment of hippocampus-dependent spatial working memory but, at the same time, display unimpaired hippocampus-dependent spatial reference memory. The GluR-A−/− mice are as impaired as hippocampal-lesioned mice on spatial working memory tasks, but at the same time, they are as proficient as control animals on tests of spatial reference memory. These findings suggest that distinct molecular pathways support different aspects of hippocampal information processing. They also pose a conundrum: Why does an impaired spatial working memory system not impinge upon spatial reference memory performance? For example, why does a memory of what happened on the previous trial not aid the mice on the present trial during spatial reference memory acquisition on the radial maze or Y-maze? Similarly, why does a memory of where the animal has just searched not help adventitious roots be more efficient in finding the platform during spatial reference memory performance in the fixed-location, hidden platform, watermaze task? Any psychological account of the role of GluR-A in hippocampal function not only needs to explain the spatial working memory impairment in the GluR-A−/− mice but, equally importantly, must also explain why this impairment has no consequences for spatial reference memory acquisition or performance.
We have previously suggested that the pattern of impaired spatial working memory but spared spatial reference memory observed with GluR-A−/− mice could be accounted for by an increased susceptibility to within-session proactive interference (Schmitt et al., 2004a; Bannerman et al., 2006). Proactive interference arises when previous reinforcement or response histories conflict with the response requirements of the current trial. An increased susceptibility to proactive interference would be expected to result in impairments on spatial working memory tasks where responding is governed by information relevant to only one trial. For example, during spatial non-matching to place (rewarded alternation) testing on the T-maze, a memory of a previous trial's different sample or choice response might interfere with memory for the present sample, or with the generation of the appropriate choice of response to it. However, in contrast, no impairment would be expected on spatial reference memory tasks where previous and current schedules of reinforcement or responding are congruent. Any tendency to recall correct responses on previous trials would simply increase the likelihood of making another, identical, correct response. We have previously suggested that GluR-A-dependent processing may contribute to reducing interference from trial to trial on working memory tasks (Schmitt et al., 2004a).