فراموشی میدازولام و بازیابی از حافظه معنایی: توسعه روش هایی برای آزمایش نظریه های حافظه ضمنی

Studies of organic anterograde amnesia have been central to the development of theories of implicit memory. Pharmacological amnesia provides an additional method for exploring implicit memory, allowing for the experimental manipulation of amnesia and the testing of more participants. A significant concern with pharmacological amnesia is whether its cognitive effects are specific to explicit memory. The current research examines the effects of the benzodiazepine, midazolam, on retrieval from semantic memory and encoding in explicit memory. We focus on midazolam because it holds significant advantages over other benzodiazepines in inducing pharmacological amnesia and prior research suggests it may be useful for testing theories of implicit memory. Our results demonstrate that midazolam does not impair accuracy of retrieval from semantic categories, even when it produces anterograde amnesia for retrieved category items on a later recall test. These results suggest ways midazolam can be used to help test theories of implicit memory.

Studies of organic amnesia have been central to the development of theories of implicit memory (Blaxton, 1995; Graf, Shimamura, & Squire, 1985; Graf, Squire, & Mandler, 1984; Schacter, 1987). Demonstrations of preserved memory capacities in organic amnesiacs have suggested the existence of multiple forms of memory, with the distinction between explicit (i.e., conscious) memory and implicit (i.e., unconscious) memory being central to contemporary approaches. Pharmacological amnesia (Curran & Birch, 1991; Ghoneim & Mewaldt, 1990; Hennessy, Kirkby, & Montgomery, 1991; Hirshman, Passannante, & Arndt, 1999; Mintzer & Griffiths, 2001) provides an alternative methodology for testing hypotheses about implicit memory. In studies of pharmacological amnesia, participants are administered a drug that induces a brief anterograde amnesia and their memory is tested. The performance of these participants can be compared to their performance in a placebo condition to understand the effects of pharmacological amnesia on explicit and implicit memory. Studies of pharmacological amnesia have important methodological advantages over studies of organic amnesia. There are a larger number of potential participants, enhancing opportunities to replicate findings and amnesia can be experimentally manipulated, mitigating problems that arise because of the correlational nature of organic amnesia. Moreover, participants can act as their own controls when administered a placebo. Coupled with the larger number of participants, this substantially increases the statistical power to detect experimental effects. As a further contrast to organic amnesia, amnesia-inducing drugs can be administered at various times during an experiment, making it possible to identify the stage in processing (i.e., encoding/consolidation or retrieval) that mediates experimental effects. A significant concern in using pharmacological amnesia is whether an amnesiac agent produces specific effects on explicit memory or general effects on a range of cognitive processes. For many drugs (e.g., fentanyl, Veselis, Reinsel, Feschenko, & Wronski, 1997), the dosage necessary to produce substantial anterograde amnesia will produce extreme sedation, significantly impairing attention, short-term memory, retrieval from semantic memory and implicit memory. Drugs that produce such general patterns of cognitive impairments have limited uses in the exploration of theories of implicit memory. The purpose of the experiment presented here is to explore the cognitive specificity of the benzodiazepine, midazolam (Curran & Birch, 1991; Evans & Viola-McCabe, 1996; Hennessy et al., 1991; Hirshman, Passannante, & Arndt, 2001). Midazolam is a safe, fast-acting benzodiazepine used in anesthesiology (e.g., surgical pre-medication) and psychiatry (e.g., treatment of anxiety). Specifically, we focus on the question (Tulving, 1984) of whether midazolam spares retrieval from semantic memory, while it impairs encoding in explicit memory. We investigate midazolam because it may have significant methodological advantages over other benzodiazepines in producing pharmacological amnesia. Because it is water soluble, midazolam is metabolized substantially faster than other benzodiazepines such as diazepam and lorazepam (Stoelting, 1991). This rapid metabolism is convenient for investigations of implicit and explicit memory because it allows one to examine test performance at relatively brief retention intervals (e.g., 1 h) with the assurance that performance on baseline or new items will be equal in the midazolam and placebo conditions. This assurance mitigates concerns about scale effects that often arise when baseline or new performance is substantially lower in an amnesiac than a control condition (Ostergaard, 1998). Similarly, the amnesiac effect of midazolam has a more rapid onset than that of other benzodiazepines (Stoelting, 1991). This rapid onset permits a much shorter interval between the administration of midazolam and the study period, shortening the length of experiments substantially. Midazolam’s water solubility is also important to its use in a non-clinical population because water solubility minimizes the side effects of intravenous administration. Intravenous injection of midazolam, in contrast to intravenous injection of non-water soluble benzodiazepines, is painless and is unlikely to cause venuous irritation or thrombophlebitis (Stoelting, 1991). Motivation for examining the specific question of whether midazolam spares retrieval from semantic memory arises from our prior investigations of midazolam’s effect on explicit and implicit memory (Arndt, Passannante, & Hirshman, in press; Hirshman et al., 1999; Hirshman, Passannante, & Henzler, 1999; Hirshman et al., 2001; Hirshman, Fisher, Henthorn, Arndt, & Passannante, 2002). The cited studies have demonstrated that midazolam has substantial effects on encoding in explicit memory, but has limited effects on implicit memory performance. For example, Hirshman et al. (2001) demonstrated that midazolam administered prior to the study period substantially diminished generation effects on the explicit memory tests of free and cued recall, as well as overall performance on these tests, but had no detectable effect on generation effects found in the implicit memory test of cued perceptual identification. Such findings suggest that midazolam can be used to examine implicit memory test performance that is relatively uncontaminated by the effect of explicit memory processes. This novel use of midazolam holds great promise for exploring prominent theories of implicit and explicit memory (e.g., Roediger & Blaxton, 1987; Roediger & Weldon, 1987; Schacter et al., 1994 and Schacter et al., 1995) if it can be demonstrated that administration of midazolam prior to the study period spares retrieval from semantic memory during this period. Specifically, the cited theories assert that semantic processing during the study period should influence performance on explicit memory tests, but not on implicit memory tests (see Richardson-Klavehn & Gardiner (1998) for a review). It has been difficult to test this assertion using normal subjects in the past because results (e.g., Masson & MacLeod, 1992) demonstrating effects of semantic processing during study on an implicit memory test may have been attributable to contamination by explicit memory (see Jacoby (1991) for a discussion of these methodological issues). Thus, if midazolam spares semantic retrieval during the study period, we can use it to test whether semantic processing during study can influence implicit memory, independently of explicit memory contamination. In this context, the current experiment is part of a larger research enterprise (e.g., Jacoby, 1998) dedicated to developing methods to test theories of implicit memory. Empirical research on other benzodiazepines suggests that midazolam may spare retrieval from semantic memory. Other benzodiazepines, such as diazepam and lorazepam, impair encoding in explicit memory, but produce minimal effects on retrieval from semantic memory (Curran, Schiwy, & Lader, 1987; Danion, Zimmerman, Willard-Schroeder, Grange, & Singer, 1989; Fang, Hinrichs, & Ghoneim, 1987; Ghoneim & Mewaldt, 1975; Green, McElholm, & King, 1996; Hinrichs, Mewaldt, Ghoneim, & Berie, 1982; Lister, 1985; Lister & File, 1984; Mewaldt, Hinrichs, & Ghoneim, 1983). While these effects do not guarantee that midazolam will spare retrieval from semantic memory (especially given the difference in the time course of effects between midazolam and other benzodiazepines), the common physiological mechanisms of the benzodiazepines suggest the possibility of similar cognitive effects. Specifically, benzodiazepine receptors are concentrated in the hippocampus and related regions (Kraemer et al., 1995; Mohler & Richards, 1983) known to contribute to performance on explicit memory tests (Squire, 1992). By binding to hippocampal benzodiazepine receptor sites, benzodiazepines facilitate the action of γ-aminobutyric acid (GABA), an inhibitory neurotransmitter (Hindmarch & Ott, 1988). This facilitation of GABA is hypothesized to inhibit the functioning of the hippocampus and related regions, inducing temporary amnesia. Given these shared physiological mechanisms, it is plausible to hypothesize that midazolam, like other benzodiazepines, will produce effects primarily on explicit memory, sparing retrieval from semantic memory. Additional findings in the prior literature reinforce the suggestion that midazolam’s effects may be specific to encoding in explicit memory. Veselis et al. (1997) demonstrated that midazolam produced substantially more amnesia than an equi-sedative dose of fentanyl (a common sedative). This result suggests that midazolam amnesia is largely independent of the mild sedation midazolam produces. Similarly, Polster, McCarthy, O’Sullivan, Gray, and Park (1993) examined the effect of injecting midazolam prior to study and/or test. They found that midazolam only produced amnesia when injected prior to study. Moreover, there was no evidence that injecting midazolam prior to the memory test reduced the amnesia produced by injecting midazolam prior to study. The former finding suggests that midazolam produces a selective effect on encoding in explicit memory, while the latter finding suggests that midazolam amnesia is not a state-dependent effect. In summary, the current study examined midazolam’s effects on semantic processing, as well as explicit memory. As discussed above, questions regarding the effects of semantic processing during the study period on memory have been central in recent theories of explicit and implicit memory (e.g., Richardson-Klavehn & Gardiner, 1998). As such, preserved semantic processing following midazolam administration is critical if midazolam amnesia is to be used to examine the effects of semantic processing during study on implicit memory. We examined the effect of midazolam on retrieval from semantic categories. We compared the number of category members participants retrieved in 3 min following injection with midazolam and a saline placebo. In addition, we asked participants to recall the items retrieved from semantic memory on a later test. The purpose of this recall test was to verify that the current dose of midazolam produces amnesia in explicit memory. There is currently little evidence on midazolam’s effect on semantic retrieval. Hirshman et al. (2001) demonstrated that participants could accurately generate weak semantic associates using a cue word and a fragment of the target (e.g., sugar-sw_ _t) following the administration of midazolam. Similarly, Arndt et al. (in press) demonstrated that participants could accurately generate medium-ranking category exemplars using the category name and a fragment of the target (e.g., fruit-str_wb_rry) following a midazolam injection.