Basic processes in visual perception, Sintesi del corso di Psicologia Cognitiva

Scarica Basic processes in visual perception e più Sintesi del corso in PDF di Psicologia Cognitiva solo su Docsity! CHAPTER 2 Basic processes in visual perception Introduction The specific processes we use in visual perception depend on what we are looking at and on our perceptual goals. One the one hand, we can sometimes perceive the gist of a natural scene extremely rapid. On the other hand, sometimes it took several second to perceived a scene accurately. Vision and the brain Considering the brain systems involved in visual perception, there are at least 30 district brain areas working for it. The visual cortex consists in the entire occipital lobe and some parts of the parietal and temporal lobes. But we have to consider first what happens between the eye and the cortex. From eye to cortex Two types of visual receptors cells in the retina: Cones (mostly in the center, used for colour vision and sharpness of vision) and Rods (about 125 million rods are concentrated in the outer regions of the retina, they are specialised for vision in dim light). Many differences between cones and rods stem from the fact that a retinal ganglion cell receives input from only a few cones but from hundreds of rods. The main pathway between the eye and the cortex is the retina-geniculate-striate pathway. It transmits info from the retina to V1 and V2 via the lateral geniculate nuclei of the thalamus. Retinopy is the term to describe the similar working pattern in both retina-geniculate-striate system and the retinal system, basically two stimuli adjacent to each other in the retinal image will also be adjacent at higher levels within that system. Thus, retinal receptors are mapped to points on the surface of the visual cortex. Each eye has its own optic nerve, they meet at the optic chiasm. The axons from the outer halves of each retina proceed towards the same hemisphere side, whereas the inner halves cross over towards the opposite hemisphere. Then signals proceed along two optic tracts, one containing signals from the left half of each eye and the other one the signals from the right half. After the optic chiasm the optic radiation proceed to the lateral geniculate nucleus. Nerves impulses finally reach V1 in primary visual cortex, before spreading out to nearby visual cortical areas such as V2. There are two relatively independent pathways within the retina-geniculate-striate: • The parvocellular pathway: it’s most sensitive to colour and to fine details; most of its input comes from cones. • The magnocellular pathway: it’s most sensitive to movement information; most of its input comes from rods. There is also a Koniocellular pathway, but its functions are still not well understood. Early visual processing: V1 and V2 Three general points: 1. The notion of receptive field: for any given neuron is the retinal region where light affects its activity; 2. Neurons often influenced each other (e.g. the lateral inhibition: reduced activity in one neuron is caused by activity in a neighbouring neuron, it increases the contrast at the edges of objects, making easier to identify the dividing line between objects); 3. Early visual processing involves large areas within the primary (V1) and secondary (V2) visual cortex. Two pathways 1. The P pathways associates with the ventral pathway that proceeds to the inferotemporal cortex (the ventral or “what” pathway is mainly concerned with form and colour processing and object recognition); 2. The M pathways associates with the dorsal pathway that proceeds to the posterior parietal cortex (the dorsal or “how” pathway is more concerned with motion processing). There are extensive interactions between the two pathways. Organisation of the visual brain V3: form processing; V4: colour processing; V5/MT: motion processing. The ventral stream includes V1, V2, V3, V4 and the inferotemporal cortex. The dorsal stream includes V1 via V3 and MT (medial temporal cortex) to MST (medial superior temporal cortex). Three important points: 1. There are complex interconnections among visual cortical areas; 2. The brain areas within the ventral pathway are more than twice as large as those within the dorsal pathway; 3. Cells in the lateral geniculate nucleus respond fastest when a visual stimulus is presented followed by activation of cells in V1. However, cells are activated in patient with akinetopsia (TD) showed a severely impairment in perceiving direction of a high-speed visual motion but not low-speed motion, suggesting V5 is less important for processing low-speed than high speed motion. V5 is not the only one involved in motion processing, also MST seems to be involved. Some patients with MST area damaged showed problems relating to motion perception (e.g. bumping into walking people). These findings suggest MST is involved in visual guidance of walking. The notions that motion perceptions depends almost exclusively on V5/MT and MST and that those areas only process info relating to motion are both oversimplifications for various reason: • Several areas outside these two are involve in motor perception; • Approximately 60% of cells within V5/MT respond to binocular disparity and 50% of cells within V5/MT respond to stimulus orientation. However, V5/MT is especially important with respect to direction of motion with approximately 90% of cells responding; • We should distinguish between different types of motion perception: 1) first-order displays, the moving shape differs in luminance from its background. 2) second-order displays, there is no difference in luminance between the moving shape and the background. In everyday life, we encounter second-order displays infrequently (e.g., movements of grass in a field caused by the wind). Some patients have intact first-order motion perception but impaired second-order, whereas other show the opposite pattern. Thus, all forms of motion perception do not involve similar underlying processes. Finally, many studies emphasised the close link between visual motion processing and perception and the dorsal (“how”) pathway. So, it is of major importance. However, the ventral (“what”) pathway is also involved in motion perception. Binding problem Zecki’s theoretical approach poses the binding problem. One aspect of this problem is that object-related processing in different visual areas ends at different times, thus making harder to understand how the brain integrates these outputs in visual perception. There may be continuous integration of info starting during early stages of visual processing. Seymour et al. presented observers with red or green dots rotating clockwise and counter-clockwise. They found that colour-motion conjunctions were processed in several brain areas including V1, V2, V3, V3A/B, V4 and V5/MT+, and the binding occur as early as V2. Ghose and Ts’o reviewed research indicating progressively more integration of different kind of info (and thus less functional specialisation) during visual processing. So far, we’ve focused on increases in integration as processing proceeds from early to late visual areas. However, conscious visual perception depends crucially on recurrent processing (feedback from higher to lower visual brain areas). Observers’ expectations influence recurrent processing and it’s arguable that expectations facilitate the binding or integration of different kinds of visual information. The binding-by-synchrony hypothesis provides a solution, according to this hypothesis, detectors responding to features of a single object fire in synchrony. Of relevance, widespread synchronization of neural activity is associated with conscious visual awareness. Anyway, the synchrony hypothesis is oversimplified (why and how it happens?). The fact that a visual object processing occurs in widely distributed areas makes it implausible that precise synchrony could be achieved. Finally, note there are various binding problem: How visual features are bound together? How we bind together info over successive eye movements to perceive a stable visual world. Considering the broader context is clear that several lines of research are relevant: the gestaltists put forward several laws describing how perception organization works, research on visual search (detecting target stimulus among distractors) is also relevant. Definitely this research shows the importance of selective attention in combining features close together in time and space. Evaluation Zecki’s assumption that motion processing typically proceeds somewhat independently of other types of visual process has reasonable empirical support. However, what are le limitations? • The brain areas involved in visual processing are less specialised than implied theoretically. Cells in several areas respond to orientation, disparity and colour. Specialization was found only with respect to responsiveness to direction of stimulus motion in MT; • The visual brain area is more complex than assumed by Zecki. There are far more brain areas devoted to visual processing and each one has connections to numerous other areas; • The importance of brain network is de-emphasised and also the key role played by recurrent processing; • The binding problem has not been solved. However, integrated visual perception undoubtedly depends on both bottom-up and top- down processes. TWO VISUAL SYSTEM: PERCEPTION-ACTION MODEL Milner and Goodale argued in their “perception-action model” that there are two visual systems each fulfilling a different function. • There is a vision-for-perception system based on the ventral stream. It is used to identify objects. • There is the vision-for-action system based on the dorsal stream. It is used for visually guided action. 4 major differences between the two processing streams: 1. The V.S. underlies vision for perception whereas the D.S. underlies vision for action; 2. There is allocentric coding in the V.S., but egocentric coding in the D.S. 3. Representations in the V.S. are sustained over time whereas those in the D.S. are short-lasting. 4. Processing in the V.S. generally (not always) leads to conscious awareness, whereas processing in the D.S. does not. Two other differences have been suggested: processing in the D.S. is faster, V.S. processing depends more on input from the fovea. The two authors have increasingly accepted that the two streams often interact: “the influence of the ventral on dorsal stream processing seems to carry visual and semantic complexity, thereby allowing us to bring meaning to our action”. Findings: brain-damaged patients In patients with damage to the dorsal pathway (intact vision for perception but impaired vision for action) and those who have damage to the ventral pathway (intact vision for action but impaired vision for perception) we should observe the opposite patterns (double dissociations). Optic ataxia Patients with damage to the posterior parietal cortex. Some evidence suggests patients with optic ataxia are poor at making precise visually guided movements although their vision and ability to move their arms are reasonably intact. This kind of patients do not all conform to this simple pattern: • Somewhat different regions of posterior parietal cortex are associated with reaching and grasping movements, and some patients have greater problems with one type of movement than the other; • It’s oversimplified to assume patients have intact visual perception but impaired visually guided action. Pisella et al. obtained much less evidence for impaired visually guided action in central compared to peripheral vision (this patient can drive effectively). • Patients with optic ataxia have some impairment in vision for perception (especially in peripheral vision). Thus, these patients have difficulties in combining info from the dorsal and ventral pathway. Dorsal stream: conscious awareness According to the two systems approach, ventral stream processing is generally accessible to consciousness whereas dorsal stream processing is not (e.g. ventral stream is often involved in motor planning). There is some support for that, but, as we’ll see, recent evidence mostly provides contrary evidence. Ludwig et al. assessed the involvement of V.S. and D.S. in conscious visual perception, they manipulate the visibility of a visual target presented to one eye by varying the extent to which continuous flash suppression (rapidly changing stimuli presented to the other eye) impaired the processing of the targets. There are two main findings: • a tight coupling between visual awareness of target stimuli and V.S. processing; • a much looser coupling between target awareness and the D.S. processing. The first finding is in accord with the theory, but the second one suggests dorsal processing is more relevant to conscious visual perception than previously assumed. Furthermore, according to the perception-action model, this manipulation (continuous flash suppression) preventing conscious perception should nevertheless permit more processing in the D.S. than in the V.S., but neuroimaging studies have obtained no evidence of such pattern. Two pathways: update The perception-action model was originally proposed before neuroimaging and other techniques had clearly indicated the great complexity of the brain networks involved in perception and action. Recent researches led to developments of the p-a model in two main ways: • consideration of the various interaction between the two streams; • there are more than two visual processing streams; Many evidence from different studies: • first of all, the V.S. is often involved in visually guided action; • the D.S. is involved in visual object recognition (patients with damage to the V.S. often retain some sensitivity to 3D objects representation); • TMS applied to the posterior parietal cortex disrupts the holistic processing of faces; • the V.S. and D.S. are both sensitive to shape. So, how many visual processing streams are there? There is evidence that actions towards objects depend on two partially separate dorsal streams: • a dorso-dorsal stream (the “grasp” stream) used to grasp objects rapidly; • a ventro-dorsal stream that makes use of memorised object knowledge and operates more slowly than the other. Other evidence, coming from an investigation of the connectivity patterns among 22 visual areas, showed how these areas are probably organised within three visual streams: 1) Dorsal; 2) Ventral; 3) lateral. It has been speculated that the third one could be involved in incorporation of different aspect like vision, action and language. Overall evaluation The central assumption that there are two visual system is partially correct, even if it has received inconsistent support from brain-damaged patients and this is one of the main limits. Which are the others? • Findings based on visual illusions provide only partial support, they generally indicate that illusory effects are greater with perceptual judgements than actions but there are many exceptions. • The model exaggerates the independence of the two visual system (e.g. sensitivity to 3D objects); • The prevailing evidence suggest that the interactions between the two system is the norm rather than the exception; • The notion that there are only two visual processing streams is an oversimplification. Chapter’s key words Retinal ganglion cells Retinal cells providing the output signal from the retina. Retinopy The notion that there is mapping between receptor cells in the retina and points on the surface of the visual cortex. Receptive field The region of the retina in which light influences the activity of a particular neuron. Lateral inhibition Reduction of activity in one neuron caused by activity in a neighbouring neuron. Achromatopsia A condition caused by brain damage in which there is very limited colour perception but form and motion perception are relatively intact. Akinetopsia A brain-damaged condition in which motion perception is severely impaired even though stationary objects are perceived reasonably well. Binding problem The issue of integrating different types of information to produce coherent visual perception. Ventral stream The part of the visual processing system involved in object perception and recognition and the formation of perceptual representations. Dorsal stream The part of the visual processing system most involved in visually guided action. Allocentric coding Visual or spatial coding of objects relative to each other. Egocentric coding Visual or spatial coding dependent on the position of the observer’s body. Optic ataxia A condition in which there are problems making visually guided movements in spite of reasonably intact visual perception. Visual form agnosia A condition in which there are severe problems in shape perception (what an object is) but apparently reasonable ability to produce accurate visually guided actions. Hollow-face illusion A concave face mask is misperceived as a normal face when viewed from several feet away. Proprioception An individual’s awareness of the position and orientation of parts of their body.