29conceito de memória de trabalho refere-se ao sistema de armazenamento temporário e de processamento de informações envolvido nas diversas atividades cognitivas complexas que realizamos. De acordo com seu modelo mais utilizado, a memória de trabalho é formada por um processador com capacidade de atenção limitada, o executivo central, que controla três subsistemas armazenadores. Informações de natureza verbal são armazenadas pelo laço fonológico, um sistema duplo formado por um armazenador passivo baseado em códigos acústico-fonológicos e por um sistema ativo de recitação que impede o decaimento desses códigos. Informações referentes aos objetos e às relações espaciais entre eles são armazenadas pelo esboço visuoespacial, também considerado um sistema duplo formado por um armazenador visual e por um sistema ativo, responsável por manter informações visuoespaciais (Logie, 1995). Por fim, o buffer episódico proporciona uma interface entre os armazenadores verbal e visuoespacial, integrando e armazenando em representações complexas as informações provenientes dos subsistemas e da memória de longo prazo (Baddeley, 2000;.A divisão da memória de trabalho em subcomponentes especializados baseia-se em evidências empíricas que incluem déficits específicos em pacientes com lesões cerebrais (Della Sala; Logie, 1999), por diferentes taxas de desenvolvimento da memória verbal e visuoespacial observadas em estudos com crianças (Hitch, 1990;Logie;Pearson, 1997) e pelos efeitos de interferência seletiva no armazenamento de informações de diferentes modalidades (Baddeley, 1986; Logie, 1995). Estudos centrados na dissociação dupla também foram importantes na fragmentação da memória de trabalho. Esse método permite supor que uma dada função A é separada ou dissociada de uma função B se existirem pessoas com deficiências na função A mas função B preservada, e pessoas com a função A intacta e deficiências na função B (Shallice, 1988). Por exemplo, De Renzi e Nichelli (1975) identificaram dois grupos de pacientes com lesões cerebrais, um grupo com déficit de memória verbal e memória espacial preservada, e o outro grupo com o padrão inverso, memória verbal intacta e déficit de memória espacial. Essa dissociação dupla, confirmada em outros estudos (por exemplo, Basso et al., 1982;Hanley et al. 1991; Logie, 1986), suporta a subdivisão entre dois sistemas de armazenamento temporário, um verbal e outro visuoespacial.

Dynamic visual noise (DVN) has been shown to interfere with short-term memory (STM) based on visual imagery, but DVN interference has been difficult to observe in STM tasks that are based on visual stimuli. The present study investigated the effects of DVN and static visual noise (SVN) in a matrix pattern recall task. The participants were required to replicate a matrix pattern that was presented for 400 ms, followed by a retention interval of 9 s that consisted of DVN, SVN, or a blank screen. The results showed that both DVN and SVN impaired the recall of matrix patterns compared with the blank screen condition, suggesting that irrelevant sensory input indeed impairs information that is held in visuospatial working memory. These results support the hypothesis of a passive visual storage system (i.e., visual buffer) that is susceptible to sensory input. The results of the visual imagery and recognition tasks can be extended to cases of visual recall. The results are discussed from the perspective of methodological differences concerning experimental procedures and theoretical models of visuospatial working memory.

Logie's visuospatial memory model coherently integrates a large amount of experimental data, however, it has difficulties explaining the effects of irrelevant visual information, such as Dynamic Visual Noise (DVN). DVN interferes with memories created from mental images, but has less consistent effects on visual memory tasks. One assumption for the lack of DVN effect on visual memory is that the visual stimuli are initially coded, for a short time, at a pre-semantic visual memory and then stored in a semantic memory more stable, not accessible to irrelevant visual information. We evaluated DVN effect on performance in memory tasks using stimuli with different nameability levels. Our assumption was that most readily nameable stimuli would be faster encoded in semantic terms, and therefore would be less time exposed to the DVN effects. Visual Patterns Test matrices, classified according to nameability, were used as stimuli in recognition (Exp. 1), recall (Exp. 2) and recall based on verbal cues (Exp.3) memory tasks. DVN effect was contrasted with Static Visual Noise (SVN) effect in Exp. 1 and to a situation without noise in Exps. 2 and 3. Memory load, estimated by the complexity of stored matrices, was manipulated in Exps. 1 and 2. In Exp. 1 DVN impairs performance only with low nameability stimuli. At Exp. 2 the noise is more damaging with low nameability stimulus and the performance is the same in trials with DVN and SVN. The interaction between interference type and memory load shows that the noise interferes more in trials with greater memory load. Our results suggest, in methodological terms, that irrelevant visual interference techniques, both DVN and SVN, although useful in visuospatial memory study, have some issues that remain to be better determined, as well as the role of visual stimuli nameability. In terms of visuospatial working memory structure our results suggest the need for a component in which visual stimuli are encoded in a presemantic visual memory accessible to environmental stimuli, thus the need for a visual buffer.

The present study psychophysically investigated the laterality of low spatial frequencies (LSFs) and high spatial frequencies (HSFs) during face recognition at different exposure times. Spatial frequency–filtered faces were presented in a divided visual field at high and low temporal constraints in 2 tasks: face recognition (Experiment 1) and face gender recognition (Experiment 2). Both experiments revealed general primacy in the recognition of LSF over HSF faces. In Experiment 1, LSF and HSF facial information was more efficiently processed in the right and left hemispheres, respectively, and exposure time had no effect. Experiment 2 showed right hemisphere asymmetry for LSF faces at a low temporal constraint. These results suggest that the spatial frequency processing for face recognition is lateralized in the brain hemispheres. However, the contributions of LSFs and HSFs depend on the task and exposure time.

No abstract