Selective Dissemination

Abstract

memory of aversive information, where the MRF is involved in STM and the hippocampus is involved in LTM. (It is assumed that STM is being measured 64 seconds after an FS and that LTM is being measured 24 hours after an FS.) That stimulation of the hippocampus after trials disrupts LTM in rats is consistent with similar results for cats (5). Thus, with the use of brain stimulation after trials it is possible to dissociate SIM from LLTM, a result that suggests that STM and LTM are operating independently. Further support for this view is provided by Milner (2) 19 male Long-Evans rats, divided into groups for MRF stimulation (N = 6), hippocampal stimulation (N = 7), and no stimulation (N = 6). The animals were prepared surgically, trained, and given the FS as in the earlier experiment. The animals received 5 seconds of brain stimulation 196 seconds after the FS; and 55 seconds after the offset of stimulation they were retested for 10 minutes for retention of *the FS.The suppression ratios for brain stimulation 196 seconds after FS are shown in Fig. 2, with corresponding data (from the earlier experiment) for brain 434 stimulation 4 seconds after FS. For both groups the interval between brain stimulation and retest was the same (55 seconds). In the control group there was greater suppression at 256 seconds after the FS than at 64 seconds after the FS. However, at the 256-second retest (1) has presented an interesting and complete account of the implementation of a system for the selective dissemination of information (SDI) for cancer-related literature. Unfortunately, enthusiasm for his own approach -use of an enumerative classification -has led him into some rather sweeping claims regarding the superiority of this method over a whole host of others, which he lumps under the general heading "keyword-based" and treats in a somewhat cavalier and irresponsible fashion. Among the "keyword-based" systems are those using uncontrolled keywords (humanly assigned) and those based on subject headings and thesauri, as well as systems operating on free text (that is, searching the natural language of a machine-readable text with a computer).In actual fact, a retrieval or dissemination system, if properly designed, can function effectively via any of these methods. Under a certain set of conditions one method will be preferable to another, but all can be made to work. Schneider's criticism of existing systems ("imprecise indexing," "a high level of 'noise,'" and "occasionally provide a useful item of information to users" are among statements used) is exaggerated and highly subjective. Moreover, he fails to cite a single study to justify his criticism. Indeed, he chooses to dismiss lightly the results of the ASLIB-Cranfield Project (2), the most complete study of indexing languages yet undertaken, presumably because these results do not fit his own view of the universe.Instead, Schneider refers to a recent