Figure 911

The relative proportions of correct responses when subjects repeatedly pronounced or imaged pictures or words in a free-recall task. Adapted from Paivio and Csapo (1973).

predicted dual coding, there was an statistically additive effect on recall relative to recall levels calculated for once-presented items that had been imaged or pronounced. Fourth, in contrast to these results, when a repeated word was encoded in the same way on each presentation, the massed repetitions did not produce similar additive effects (see Figure 9.11).

Interference within a .single system

Paivio's theory sees the routes taken by perception and imagery as basically the same. For example, in talking about the non-verbal system he says that it is responsible both for the cognitive task of forming visual images and the perceptual task of scene-analysis. Therefore, any findings that demonstrate interference between perceptual and imagery tasks are a source of further evidence for the theory. That is, if performance on a perceptual task is disrupted by carrying out an imagery task, and vice versa, it is likely that both tasks are using related processing components. Such interference has been found on a regular basis. For example, Segal and Fusella (1970) asked subjects to form both visual and auditory images and then asked them to perform a visual or auditory detection task. They found that auditory images interfered more with the detection of auditory signals and visual images interfered more with visual detection. As there was some interference in all conditions it seems reasonable to conclude that there is a generalised effect of mental imagery on perceptual sensitivity in addition to a large modality-specific effect.

However, it is not enough to simply demonstrate interference. One needs to pin-point the specific processes that are responsible for the interference and also, if possible, to show how perceptual and image-based processing differ. More detailed evidence of this sort has been found in a task used by Baddeley, Grant, Wight, and Thomson (1975). In this experiment, subjects listened to a description of locations of digits within a matrix and were then asked to reproduce the matrix. The description was either hard or easy to visualise. The interfering task involved a pursuit rotor (i.e., visually tracking a light moving along a

BB Pronouncing Imaging

BB Pronouncing Imaging

Picture Word

circular track). This task results in a distinct type of interference—performance on easily visualised messages is retarded, while the non-visualisable message is unaffected—but the interference is not due specifically to the perceptual processes involved in vision. Baddeley and Lieberman (1980) have shown that if the concurrent task is specifically visual (e.g., the judgement of brightness), rather than visual and spatial (as the pursuit rotor seems to be) then the interference effects disappear. Similarly, when the concurrent task is purely spatial (i.e., when blindfolded subjects were asked to point at a moving pendulum on the basis of auditory feedback) the pattern of interference found reproduces the effects found in the original Baddeley et al. experiment. In summary, it appears that the recall of visualisable (or easily imagined) messages of the kind used in these experiments is interfered with by spatial processing rather than by visual processing, indicating that these spatial processes are somehow shared by perceptual and image-based processing within the non-verbal system (see Logie & Baddeley, 1989, for a review). However, there are also cases in which interference from purely visual processing can be achieved (see Richardson, 1999, and Wexler et al., 1998, for recent work).

These experiments show that Paivio's interference predictions really rest on the assumption that visual imagery involves visual rather than spatial representations. However, Farah, Hammond, Levine, and Calvanio (1988) have suggested that it is a mistake to argue that imagery is either visual or spatial. Rather they have shown, using neuropsychological evidence, that imagery is both visual and spatial and taps into distinct visual and spatial representations.

Neuropsychological evidence for dual coding

A natural question that arises about Paivio's theory is whether there is neuropsychological evidence for the localisation of the two symbolic systems within the brain. For instance, for most people the left hemisphere is implicated in tasks that involve the processing of verbal material. In contrast, the right hemisphere tends to be used in tasks that are of a non-verbal nature (e.g., face identification, memory for faces, and recognising non-verbal sounds). Furthermore, within each hemisphere there seems to be some localisation for the sensorimotor sub-systems: visual, auditory, and tactile (see Cohen, 1983). While dual-coding theory posits distinct symbolic systems, Paivio does not maintain that these distinct systems reside in distinct hemispheres, although the systems are localised to some extent (for evidence against this view see, e.g., Zaidel, 1976).

There is some evidence for localisation differences on concrete and abstract words that disrupt a simple left-right division. Word recognition studies, using tachistoscopes, have shown that there are hemispheric differences in the processing of concrete and abstract words (see Paivio, 1986, Chapter 12; Johnson, Paivio, & Clark, 1996). Typically, abstract words that are presented to the right-visual-field, and hence are processed by the left hemisphere, are recognised more often than those presented to the left-visual-field (i.e., processed by the right hemisphere). However, concrete words are recognised equally well irrespective of the visual field (and hence the hemisphere) to which they are presented. It should be pointed out that these findings have not been consistent, although there is a tendency for the performance asymmetries to be less consistent for concrete than for abstract words (Boles, 1983). More recently, detailed studies using event-related potentials and fMRI have confirmed many aspects of these proposals (see Holcomb, Kounios, Anderson, & West, 1999; Kiehl et al., 1999).

Converging evidence also comes from so-called deep-dyslexic patients, who have widespread lesions in the left hemisphere. Generally, they have greater difficulty reading abstract, low-imagery words than concrete high-imagery words (see Coltheart, Patterson, & Marshall, 1980; Paivio & te Linde, 1982). Plaut and Shallice (1993) have modelled these concrete-abstract effects by lesioning a connectionist net (see also Hinton & Shallice, 1991). However, the effects were modelled by representing concrete concepts with more features than abstract concepts, rather than using imagery representations. Tyler and Moss's (1997) results present some problems for this proposal because they have found a patient with a selective problem understanding the meaning of abstract words in a specific modality (i.e., auditory modality). We shall see next, in the presentation of Kosslyn's theory, that some more recent evidence presents a clearer picture for what might be happening in both hemispheres (see Farah, 1984; Kosslyn, 1987).

KOSSLYN'S COMPUTATIONAL MODEL OF IMAGERY

The work of Stephen Kosslyn and his associates tested and developed a theory that can be viewed as a response to the early criticisms of imagery theory. More recently, Kosslyn has made a strong claim for the overlap between the processes of visual perception and imagery (see also Chapter 4). In his 1994 book, Image & Brain, Kosslyn lays out a full theory of visual perception which he then maps onto his earlier 1980 theory of imagery. For the most part, the processes he originally proposed to account for imagery are now re-used (with some minor modifications) to deal with perception too. An important part of this work is the research on the neurological basis of both abilities (see later section on Neuropsychology). For simplicity's sake, we summarise the 1980 theory here (see Kosslyn, 1994, Chapter 11, for more detail).

The theory and model

Kosslyn's theory has been specified in a computational model and is roughly summarised in Panel 9.3 (see also Figure 9.12; Kosslyn, 1980, 1981, 1987, 1994; Kosslyn & Shwartz, 1977).

Consider the basic task of generating an image of a duck. The theory maintains that several structures and processes are involved: the spatial medium in which the duck is to be represented,

Business Correspondence

Business Correspondence

24 chapters on preparing to write the letter and finding the proper viewpoint how to open the letter, present the proposition convincingly, make an effective close how to acquire a forceful style and inject originality how to adapt selling appeal to different prospects and get orders by letter proved principles and practical schemes illustrated by extracts from 217 actual letter.

Get My Free Ebook


Post a comment