The DRM paradigm shows subjects a list of words and subjects falsely recall a closely related critical lure. It is assumed that at a semantic level of representation activation spreads towards the closely related word, therefore causing false recall, because the lure is also semantically activated. However, items that were truly presented are also activated on a sensory level of representation. Therefore, if you can distinguish between words that you only have semantic representation activation and words that you have both semantic and sensory representation activation then you should be able to reduce falsely recalled items.

Recent findings have shown that on written recall and visual recognition tests, false recalls have been greatly reduced by presenting items visually rather than aurally. It is assumed that the visually presented items aided subjects in distinguishing between words that they had only thought about and words that they had actually seen because visual presentation provides distinctive features. Also, when words are presented aurally and visually simultaneously, false recall is reduced. Subjects may have been using visual cues to distinguish between words they had or hadn’t seen before.

One reason this may occur is because when we are presented visual word stimuli, we orthographically (this means that we map the shape and look of the word in our mind) and phonologically (mapping the sounds) encode it. But we don’t do both with words that we only hear. We only phonologically encode words we only hear. To explain, even though we may see the word in our mind when we hear it, that image decays quickly and we are left with only phonological aspects of the word in our working memory. Therefore, we have 2 ways of distinguishing between presented and non-presented words when they are visually presented.

Furthermore, with written recall tests, both encoding processes are utilized because when we write words we refer to both our phonologic and orthographic memory for the words. This also happens with visual recall tests because you are reading each item. But if the recall test were oral, no orthographic codes would be needed so they wouldn’t be there to aid in recognition of false items.

If written recall utilizes both encoding processes at retrieval then visually presented words should be easier to recognize if the recall test is written rather than spoken. Also, would the advantage be consistent in a within-groups design? This would show that participants are using specific information to keep from falsely recalling words. In a within-groups study, the distinctiveness heuristic cannot be used for all test items because all subjects are exposed to each condition and that heuristic doesn’t work with each condition.

Subjects were presented with 4 lists that fit the DRM paradigm. One list was presented aurally, with a oral recall test, another was presented aurally with a written recall test, the other 2 were presented visually with 1 having a written recall test and 1 having an oral recall test. All subjects were exposed to all 4 conditions.

Subjects virtually cut their errors in half on the written recall tests when the words were visually presented rather than aurally presented. Nothing changed when the recall test was oral (Figure 1).

These results imply that orthographic features are retrieved when a subject is asked to write the word rather than say it, but these features are only present when the words have been visually rather than aurally encoded. Also, in the visual-written condition, the words seem to have semantic, phonological, and orthographic features being encoded, therefore, more criteria to help one distinguish new from old.

To further test whether or not orthographic features were aiding in retrieval, Kellogg wanted to make sure that aurally presented words would be encoded orthographically and see if the visually presented words kept the recall advantage. Any advantage visual presentation has should disappear if it is due to orthography when aurally presented words are orthographically encoded too. To make this happen, subjects were asked to imagine the words and report the number of curves the words had. The assumption was that subjects who were asked to imagine the words would keep the stored orthographic information available because of the task they were asked to complete (reporting number of curves).

In this experiment, the 4 conditions were aural presentation with imagery or counting and visual presentation with imagery or counting. More false memories were found with aural presentations than visual presentations when imagining was not utilized, but when it was utilized, the same amount of false recall was recorded for both presentation conditions. However, the counting conditions had more correct recalls than the imagery conditions. Perhaps all of the encoding that takes place when orthography is introduced is somewhat distracting (Figure 2).

It is suggested that the degree of encoding will affect the amount of false recall a subject has. When orthographic encoding is present as with visually presented words, false memories decrease. These experiments add support to this hypothesis by isolating orthographic representations and seeing if they are really affecting recall. This type of information provides us with distinct features to use as criteria for what will be accepted or not. Therefore, these features can aid in enhancing recognition.