A couple of comments on their findings; they're consistent with what behavior analysts know about stimulus control. To promote generalization, you should train the desired behavior in as many different situations as possible. This makes sit less likely that the behavior will occur only in the situation in which it has been taught. To teach discrimination between stimuli it is most effective to present two (or more) stimuli simultaneously, and reinforce the behavior of choosing the correct one. Serial discrimination training is also possible, but more difficult.
Paul Brandon Emeritus Professor of Psychology Minnesota State University, Mankato [email protected] On Sep 6, 2010, at 10:04 PM, Claudia Stanny wrote: > > Encoding specificity doesn't really refer to the effect of environmental > context cues, although the phenomena are related. > > Cues from context (like studying in the same room) are weak relative to > associations developed between meaning of cues and information to be > retrieved. So studying in the same physical environment is good for > short-term retrieval of weakly-learned material. When other associations > work for the retrieval and these are strong, the effects of these stronger > retrieval cues overshadow contextual cues. Contextual effects fade as the > strength of overall learning increases. > > Encoding specificity (a la Tulving and Thompson) is related to the > specificity of retrieval cues learned at the time of study and those > available at the time of test. These could be context cues, but Tulving and > Thompson focused on cues related to the material learned and the specificity > refers to the encoding of these associations. Tulving and Thompson had > students learn words in the context of weak associates as study cues (e.g., > HEAD - LIGHT). When HEAD is presented as the cue during test, retrieval was > usually successful. If a different weak associate was presented as the cue > (e.g., NIGHT), the cue was ineffective. They got this effect when either weak > associate was used as the studied cue (half studied with HEAD, half with > NIGHT). When a strong associate is presented as a free association cue > (e.g., DARK), students produce LIGHT as a free associate response, but > frequently fail to recognize this as a to-be-remembered word connected to one > of the weak associates. (A weird example of a recognition test producing > worse performance than a retrieval task.) This phenomenon supports the > advice that students should anticipate the types of questions (and types of > retrieval cues) they will get during testing and explicitly encode the > material with those retrieval cues when they study. > > A related phenomenon to this discussion is the fan effect (many studies by > John Anderson). The fan effect describes the way in which excitation for > retrieval is distributed among the many associations to a retrieval cue. The > larger the fan (the more items associated with the cue), the slower the speed > of activation and the weaker the activation of any one item. The more items > associated with a single cue, the less effective that cue will be in > retrieving any one item. However, in a semantic network in which items are > widely interconnected, this effect is minimized. That is, if an item serves > as a retrieval cue for many items, the negative effect of the fan effect is > offset by the fact that to-be-retrieved items are associated with many > potential retrieval cues. Thus, any learning that increases the number of > potential retrieval cues that might successfully activate the > to-be-remembered item will lead to more successful recall. This is why > experts retrieve knowledge in their domain of expertise quickly and > successfully rather than get mired in fan effects: Their semantic networks > are highly structured and interconnected, enabling retrieval through a > variety of routes and using various retrieval cues. This highly redundant > and interconnected associative network can emerge from distributed study > times, study in a variety of contexts, connecting new learning to other known > material, etc. All good study strategies, but effortful. > > The problem for education is that we frequently test students only after > short retention intervals. In many cases, strategies that produce good > long-term retention don't look as good as the short-term strategies in an > immediate test - the benefits of these "better" strategies only appear in > experiments in which students are tested after a delay (e.g., two weeks after > the study session). Given the way we test, students who use strategies that > enhance short-term retention at the cost of long-term retention (cramming, > relying on cues in a single environmental context) get reinforced for using > these strategies instead of strategies that produce deeper, long-term > retention but take more effort and might not produce results as good as > cramming in a short term test. --- You are currently subscribed to tips as: [email protected]. To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5&n=T&l=tips&o=4667 or send a blank email to leave-4667-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu
