Visual Memory

(source: Detailed Visual Memory for Objects in Scenes following Intentional and Incidental Learning Tasks; Monica S.Castelhano and John M.Henderson, Department of Psychology and Cognitive Science Program[sic] Michigan State University)
...Results suggest that detailed visual information is encoded regardless of task instruction...Studies of transsacadic memory have shown that visual information is not fused accross saccades...research on change detection has found that large discrepencies in a scene can go unnoticed if changes are made during a saccade...given such poor change detection performance, some researchers have abandoned the idea of a complete internal replication of the outside world. O'Regan (1992) contends that a complete visual representation is not necesary because the outside world acts as its own 'memory'. Other researchers have claimed that the representation is temporally and spacially limited and is confined solely to where attention is focused. Like other theories, Hollingworth and Henderson's visual memory theory also assumes that the repersentation in VSTM is constrained to the current and previously attended regions. It differs, however, in that the temporary can be transfered to to LTM; therefore the visual information of a previously fixated object is never completely lost once attention is deployed elsewhere. in their third experiment (2002) Hollingworth and Henderson used an online memory test and showed that participants were able to discriminate betwwen a previously fixated object and a distractor of the same basic-level catagory (eg., distinguishing between two kinds of radio). More importantly, performance was still high when participants has to discriminate between the original view of the critical object and a 90° rated view of the same object...one way of explaining these results is to posit that visual information is stored in LTM once attention is removed and that the current representation is updated accordingly...the question of whether visual details are aquired incidentaly in the course of normal scene perception is important in order to determine the extent to which visual representations are retained in LTM.[a recent experiment]suggested that once an object in a scene has been attended, visual information about that object is stored in LTM, not simply VSTM...the visual details necessary to perform the memory test must have been encoded even when there was no demand to learn those details. In light of these and past findings and the present results, it seems that visual search does indeed have memory. [Dyslexics may rely] heavily on long-term memory. LTM is based on meaning [?] so adult dyslexics often have a very personalised approach to remembering. This may involve pattern or colour or it may involve memories that relate to their personal associations. [for more information click here]...the present study strongly suggests that once an object is attended and a representation is constructed, that representation is stored in memory regardless of the intentions of the viewer or the demands of the task...many studies in the 70's and 80's of the consistency affect found that objects inconsistent with the overal schema of the scene are recognised and recalled more often than consistent objects...specifically it has been shown that attention is necesary to notice a change and change detection performance greatly increases when an object is fixated before and after a change...recent studies have found that changes made to objects that are inconsistent with a scene schema while the picture is being viewed are more likely to be detected than consistent objects...these results suggest that attention is the mediating factor in determining which information is encoded and retrieved. Future research on visual memory must therfore take into account visual and semantic factors in the stimuli that will affect how information is attended and, ultimately, encoded.
The notion of retrieval failure is consistent with other recent studies. Simons, Chabris and Schur(2002) found evidence for the retention of a pre-change representation despite change blindness. In this study participants were pedstrians giving directions to the presenter carrying an atypically coloured basketball. During the exchange a group of people passed around and between the participant and the experimenter, surreptitiously removing the basketball. (half the participants were exposed to an additional manipulation in which the ball appeared after the group went by). When asked open-ended questions about the change the majority of participants demomstrated change blindness and failed to report any change. However, when asked pointed questions about the object roughly 70% of these participantswere now able to 'discover' their memory for a basketball and able to give detailed descriptions of what the specific ball looked like. This led Simons et.al. to conclude that some representation of the pre-change scene did exist and that change blindness may be the result of the failure to retrieve and compare the past representation to the current one. Only when the appropriate memory cues are given can the comparison take place.
Although the extent to which visual information is thought to be stored varies with each theory, there does seem to be a general concensus across all theories on at least some of the characteristics of the representation used during current perceptual processes. The representation is thought to be fragmentary and centred on objects. It is believed that pre-attentive processes mark spatial locations of where objects are likely to be found and in turn, guide attentional processes there. Once an object is fixated attentional processes build up objective representations. The current visual representaion is also considerd to be limited both spatially and temporally to the information currently being attended and perhaps at best, 2 0r 3 additional objects.