Our experience with Memorin in elderly subjects diagnosed with age-related cognitive decline (DSM-IV) has been very encouraging. Memorin capsules administered four per day in two divided doses produced significant improvements in several measures of memory relative to placebo (Andrade et al., 1998). A small pilot study comparing Memorin and placebo in patients receiving ECT has recently been completed, and the data are presently under analysis.

Shankapushpi

Shankapushpi (Evolvulus alsinoides) is an ingredient of both Mentat and Memorin. Shankapushpi is highly rated in Ayurveda as a treatment for impairments related to the central nervous system. Accordingly, we examined the ability of an aqueous extract of Shankapushpi to promote learning, and to attenuate ECS-induced anterograde and retrograde amnesia studied using the T maze and the Hebb-Williams maze.

The results were altogether disappointing. Shankapushpi did not enhance learning performance on either task, nor did it attenuate either anterograde or retrograde amnesia induced by various schedules of ECS in rats (abstracted in Andrade et al., 1996). It is of course conceivable that Shankapushpi may contain a procognitive ingredient that does not emerge in an aqueous extract; if so, an alcoholic extract of Shankapushpi, or extracts obtained by some other process, may yield more encouraging results; this issue will require evaluation in future experiments.

Caveats

Herbal medicines are prepared from the leaves, roots, and other parts of specific plants. The biology of these plant parts varies as a function of their location on the plant, the time of day, the season of the year, the cultivation process, variations in weather and soil, and other factors. Accordingly, standardization of an herbal pharmaceutical product requires much care.

Today, the principal chemical ingredients of most of the important herbal source materials are known and have been published (e.g., Kirtikar and Basu, 1993; 1994). What is uncertain, however, is the identity of the chemical that is biologically relevant in a particular herb. Most herbal pharmaceutical companies therefore obtain a chromatographic "fingerprint" of a gold standard of their herbs, and endeavor to ensure that all subsequent batches of their products match this fingerprint. The shortcoming of this procedure is that the standardization process may be based upon irrelevant ingredients.

Animal Models of Cognition:
Theoretical and Practical Issues

Animal models of cognition are well described in the literature and will not be reviewed here. Instead, we present certain theoretical and practical issues that arise from the conduct and interpretation of research based on such models. We focus on our own experiences in this regard, derived from the studies described in the earlier section.

General Limitations of Animal Models

Conducting research on human subjects may yield the most reliable results, but is expensive, time-consuming, and fraught with ethical difficulties. The use of animal models of physiological or psychological function or dysfunction is therefore helpful during the early stages of hypothesis generation, during drug development, and in other contexts of explorative research. The utility of animal models notwithstanding, it must be remembered that resulting findings are generalized to human contexts; this exposes the limitations of such models. Consider the following issues:

1. A rat is far removed from a human; the validity with which comparisons can be drawn between rodent and human research is therefore uncertain. For example, the complex processes described under registration, retention, recall, and recognition under short- and long-term storage conditions in humans may not apply to the same extent in rats. The neurophysiology and neurochemistry of a rat may be simpler than that of a human, making it easier for a drug to have a demonstrable procognitive effect in the former than in the latter situation. A contrary view is also conceivable: A complex human system may have more sites at which a multiingredient herbal compound could act, making the compound more likely to be effective in the human context than in the laboratory context.

2. A healthy rat is far removed from a dysfunctional human. For example, even if memory processes are identical in rats and humans, it is uncertain whether memory processes in healthy rats are similar to memory processes in humans who are modified by conditions such as depression and schizophrenia. It is likewise uncertain whether a drug that has procognitive effects in a healthy rat will have procognitive effects in a human whose biology is compromised by the neurophysiological, neurochemical, and neuroendocrine changes associated with psychiatric illness.

3. Even if healthy rats can be equated with dysfunctional humans, animal models of psychological states are still remote approximations of what they are clinically considered to represent. For example, the processes that delay a rat's ability to locate the reward chamber in the Hebb-Williams maze are likely to be much different from the processes that underlie ECT-induced autobiographical memory impairment, if only because maze learning is a spatial task while autobiographical memory is nonspatial. A drug that is effective in one context may therefore not be effective in the other context.

4. Even if animal models correspond perfectly with the human processes that they are desired to represent, the absence of internal and external "noise" in laboratory contexts prejudices the generalizability of animal studies. For example, laboratory animal~ used in research usually belong to the same age, sex, and inbred strain; they therefore closely resemble each other in behavior. Furthermore, the laboratory environments in which the animals are housed and the experiments conducted are both carefully controlled, and are kept constant all through the experiment. All these factors reduce the variance of the results in animal experiments. In contrast, in human contexts interpersonal and environmental differences across subjects are multiple and are very difficult to control. These factors increase the variance of results in clinical research. The consequence of low variances in animal research and high variances in clinical research is that statistical significance is far more easily attained in the laboratory than in the clinic. Thus, for example, a drug that has a small procognitive effect may produce statistically significant results in the laboratory and insignificant results in the real world. In other words, the small positive effect of the drug is, in human research, drowned out by the background noise. This may be one of the reasons why many procognitive treatments that have been shown to be effective in animal models prove to be ineffective in clinical trials. A point worth noting is that small positive effects, if they exist, can be demonstrated in clinical contexts if a sufficiently large sample is studied; however, it is uncertain whether the statistically significant results so obtained would be clinically meaningful.

Thus, it is necessary to generalize with caution between animal and clinical research; when such generalizations are made, the limitations of animal models must be kept in mind.