Sensation and Perception PDF

Preview

Summary

Sensation andPerceptionDevelopmentLow Level VisionYou have the sense that you’re seeing your environment directly as it is, but in reality what you are perceiving is a complex representation that reconstructs aspects of the world around you and leaves out many othersThe aspects of the physical world...

Description

Sensation and Perception Development

Low Level Vision You have the sense that you’re seeing your environment directly as it is, but in reality what you are perceiving is a complex representation that reconstructs aspects of the world around you and leaves out many others The aspects of the physical world that your nervous system reconstructs and the perception it gives you, have a huge impact on your concepts of the world and yourself in it. If you apply electrical signals to neurones in the brain’s primary sensory areas that replicated organs in the real world, then this situation is indistinguishable perceptually from reality.

Perception is something entirely constructed by the brain

Perception is a selective take on reality that depends on both the sensory inputs and also on how the brain processes them.

Aim: to understand… - the concept of receptive fields and how their structure increases in complexity through the visual processing hierarchy. - how different visual attributes are extracted and encoded independently must be reassembled to allow a coherent perception of the external world - how converging evidence from different methods is used to explore the structure and function of the visual system.

Studying Perception: Neuroimaging - MEG - fMRI - EEG - optical imaging

Electrophysiology - single electrode - multi-electrocode

Psychophysics

*** methods can be direct (imaging the brain/measuring the electrical activity of visual neutrons) OR Indirect (psychophysics present visual stimuli to ppts and ask them to make perceptual judgements and provide behavioural responses to what they are seeing). Then we can infer some of the processes the brain uses to achieve those perceptions.

The Visual System - These methods have led us to a relatively good basic understanding of how the visual system is structured: - We know that light is transduced by photoreceptors. - We know that electrical signals are sent from the retina through the lateral geniculate nucleus to the visual cortex (which is divided into many areas which are labelled according to numbering system such as V1, V2, V3.

This circuit diagram shows that the visual system is hierarchically organised and that information travels from the retina, through to the primary visual cortex V1 and then through other cortical visual areas, which splits into two broad processing streams. “Where” pathway: travels into the parietal cortex; processes aspects of object location and motion. “What” pathway: proceeds through the temporal cortex; processes aspects of object form and identity.

*** despite knowledge of basic facts about visual system, the detail of our understanding is constantly evolving - so the process is a lot more complex. When the light from an object enters the eye through the pupil, it is focused by the lend onto the retina at the back and the image of the object is cast on the retina Image of retina through an ophthalmoscope

retinal blood cells, optic disk and blind spot where the optic nerve exists the eye. Small shadow = fovea, subserves the detailed vision at the centre of your case.

zoomed in image of a patch of retina and you can see the individual photoreceptors Taken with an electron microscope and shows the rods falsely coloured in yellow and the cones falsely coloured in blue.

Large discs = cones, active in light conditions Small discs = rods, active in dark conditions, allow night vision

shows rods and cones structure, outer segments contain many folds of cell membrane In the membrane, are embedded photosensitive molecules that change their shape when they intercept light. This starts a chemical cascade, when then leads to an electrical signal at the base of the photoreceptors

This shows a transverse section through the retina, taken using a microscope. - the retina is wired backwards: the rods and cones are at the bottom, furthest from the pupil and all the other processing layers are above. The dip is at the central fovea where some of the layers are bent out of the way because that area has the highest acuity so it needs to let as much light in as possible to achieve that fine detail

This diagram shows the different cell populations that make up the basic retinal circuits The image is inverted compared to the photograph, because the rod and cones are at the top, and they connect to bipolar cells (yellow) which carry signals to the retinal ganglion cells (brown). - these are the output neutrons which send signals to the brain The horizontal cells (green) and the amacrine cells (grey) are two cell populations that process info laterally across the retina - meaning that what happens t any given bipolar cell depends both on its direct connections to rods and cones, but also what is happening at the neighbouring areas of the retina. = the retina is doing fairly complex computations!

Bipolar Cell Receptive Fields This diagram shows what happens to a bipolar cell when light hits the photoreceptors on a patch of retina. The bipolar cell makes direct connections with photoreceptors directly above (blue) Then, when the light is received by these central photoreceptors (red) These indirect connections apply the horizontal cells and they reverse the sign of the signal, so when light is received by the surrounding photoreceptors, the bipolar cell is inhibited. The whole area on the retina, which will triggers a response of some kind in the bipolar cell, is its receptive field. Receptive Field: *A neuron’s receptive field is the area on the retina which will cause a response in the neuron when photoreceptors at that area are stimulated by light The receptive field of the bipolar cell has what’s called a centre surround structure - this type of bipolar cell is called on centre, because it has two regions: Central region - causes an excitatory response + Surround region - causes an inhibitory response The on-centre bipolar cell’s optimal stimulus (the one causing the greatest response), is a bright spot of light covering the receptive field surrounded by a dark region. There are also bipolar cells that have an off centre and an on surround receptive fields if light is received by the central photoreceptors. There is an inhibitory response and if light is received by the surround photoreceptors, it’s excited. In this case, the bipolar cell’s optimal stimulus is a dark spot covering their receptive field centre, surrounded by a bright background. Bipolar cells connect to retinal ganglion cells (the retinas output neurones) and send signals to the brain. These inherit the same centre surround receptive field structure as the black powder cells.

Demonstration of what this means for perception: If you stare at the blue dot for ~ 20 secs, the fuzzy grey blob will disappear… This is because it is a very poor stimuli for the visual system. On average, the same amount of light is received by the centre and the surrounds, which means the inhibition cancels excitation and the ganglion cell doesn’t signal much at all, so the brain receives no signal that this stimulus is still there, so the blob fades from perceptual experience.

BUT… this image of a hard disk doesn’t fade easily, because any retinal ganglion cell who’s receptive field is near the edge of the image of this disc signals very strongly. The centre here is very excited by the white background, and the surround is only partly inhibited by the image of the background, and some of the background falls on the dark part of the disc, which means that the surround receives less light than the centre… So, the inhibition doesn’t cancel the excitation and the ganglion cell provides a strong signal to the brain that a stimulus is present. Take home message: the retina (or the centre surround structure of the bipolar and ganglion cells’ receptive fields in the retina), means that the retina cares about edges between dark and light regions in the image, and doesn’t care so much about uniform regions This image shows the photograph as it is signalled by on-centre retinal ganglion cells - largely represent edges/borders between darker and lighter regions This is the visual information that leaves the retina for the brain, showing hoe vision isn’t an accurate pixel-by-pixel mapping of the retinal image, but in fact the brain receives info that’s different from info contained in the photo —> Then, perception must reconstruct the physical world from quite dramatically distorted inout signals

Two Major Classes of Retinal Ganglion Cell - both have on and off centre subtypes; they signal different aspects of visual info. Midget Cells: - small receptive fields - sensitive to fine detail - not sensitive to fast moving light or flickering light - red and green selective (respond to red/green colour differences - ‘what’ pathway originates from midget cells and signals aspects of object, form and colour

Parasol Cells - larger receptive fields - only sensitive to larger scale visual stimuli and are sensitive to fast flicker and motion - not colour sensitive - ‘where’ pathways originate from parasol cells and carries info about object, motion and location.

*** these different classes of ganglion cells send different types of visual info to the brain and they represent the start of two different parallel pathways that remain separate the way through the visual processing hierarchy. *** the brain must put all the streams of info back together to enable a coherent perception of the external world - In the primate retina, there are 12 types of bipolar cell, more than 40 types of amacrine cell and more than 17 distinct types of retinal ganglion (each anatomically different, each carrying a different type of visual info to the brain)

Lateral Geniculate Nucleus - LGN - part of the thalamus - used to be considered a relay station, but It takes input from both of the yes and the connections from the eyes to the two LGNs are from halves of the retina. - info from the LVF is processed in the right half of the brain and vice versa - the separate visual pathway signalled by the midget and parasol cells remains anatomically segregated in the LGN

- signals sent from the retina, which originate in the parasol cells, are segregated into two layers in the LGN, which are part of the magnoccellular processing stream, and then which give rise to the ‘where’ pathway - whereas the signals carried by the midget cells are received in four layers of the LGN and are part of the processing stream which gives rise to the ‘what’ visual pathway.

The midget cells connect to parvocellular LGN neurons, while the parasol cells connect to magnocellular LGN neurons. - neurons in the LGN (like those in the retina and every visually sensitive neurones in the visual system) have receptive fields, whose structure is inherited from the retinal ganglion cells, and has the same centre surround organisation. - visual info travels from the LGN through the optic tradition to the primary visual cortex at the back of the brain. Before travelling forward again through other cortical processing regions

Hubel and Wiesel accidentally discovered that the structure of the receptive fields of neurons in V1 is elongated, and they preferentially respond to oriented lines or edges.

The receptive fields of neurone and V1 are constructed by wiring together inputs from neighbouring cells in the LGN, which are themselves connected to neighbouring cells in the retina —> IMPORTANT: neighbouring cells in visually responsive brain regions signal neighbouring positions on the retina and thus neighbouring positions in the visual field: retinotopy essentially visual areas in the brain are spatially organised to systematically represent the visual field from left to right and bottom to top.

The receptive field of the V1 neurone is the sum of the receptive fields of the three input cells. And the result is that if light falls in this elongated region, which is caused by wiring together the three receptive field centres, the V1 cell is excited, while it's inhibited, if light falls on the elongated surround and the V1 neurone therefore has a laterally extended receptive field and these cells are known as simple cells and are sensitive to oriented straight edges or lines.

Receptive Field Structure Increases in Complexity… - The complexity of receptive fields, visual neurones builds further as we go through the visual processing hierarchy - So in the retina, they have this centre surround organisation and then cells in V1 wire together neighbouring retinal ganglion cells to achieve receptive fields that are selective for particular orientations of line. - Then more cells of wire together to create cells that are sensitive to lines of a particular length, and a prominent idea in vision science has been that this complexity keeps building through the visual processing hierarchy so that visual neurones become so active for more and more complex patterns of light,dark and colour cast on the retina - Features of the objects would be reassembled by higher level neurones, receiving converging inputs from the all the relevant feature selective neurones at lower stages. - The pinnacle of the process would be cells that have receptive fields that are selected for particular objects. - These hypothetical cells have been called grandmother cells because they would be selective enough to respond only to an image of an observer's grandmother. Evidence: from electrophysiology that the complexity of receptive fields keeps building through the 'what’ processing pathway, also known as the ventral stream. - for example, in area TEO, which is near area IT quite far along the ventral stream, neurones have a variety of receptive fields like these

in this figure, which are fairly complex patterns of light and dark, which are needed to optimally stimulate the neurones.

Quiroga et al. made single cell recordings around the hippocampus, which is right at the end of the ventral stream in eight human participants who were undergoing brain surgery while they were awake - they showed each patient different images and different views of distinct objects and faces.

- They found that certain neurones were activated by one unique face, but virtually silence for other stimuli. - Here is an example of a neurone that seemed to be activated exclusively by different views of Jennifer Aniston. Another example. A single cell activated exclusively by different views of Halle Berry and also by the printed name Halle Berry. - So these single neurones appear to respond like grandmother cells because they are activated by the image of one unique object, but not not other objects, even in the same category. BUT… Quiroga was very cautious about calling these grandmother cells: - for one thing they also found neurones that responded to more than one stimulus. - during the session, they presented only 16 objects to each patient. - so it's statistically very unlikely that they would have hit on the stimulus preference of the grandmother cell. From their results, they actually estimated that out of one billion medial temporal lobe neurones, fewer than two million represent each given object. - So what they found isn't grandmother cells, but a sparse coding system for unique objects - So, for example, they found a neurone that seemed to respond to the concept of Luke Skywalker, and this would be one of a collection of distributed neurones in the medial temporal lobe that together form a network representing Luke Skywalker. … and that network would be connected to other networks for related concepts like Yoda and Darth Vader. - So it seems that grandmother cells where a single neurone represents a single concept probably don't exist, but perhaps single objects are represented by rather a small network of neurones.

Retinopy - neighbouring neurones in visually responsive brain areas represent stimuli at neighbouring areas of the retina. - similar to somatotopy: where neighbouring areas of primary somatosensory and primary motor cortex represent neighbouring regions of the body - and tonotopy: where in primary auditory cortex neurones representing different frequencies of sound or systematically organised.

This diagram here shows that in the central figure where the BOLD response is shown in red is represented right at the back of V1 and then the peripheral visual field is systematically represented further forwards or anteriorly.

In vision, the organisation is spatial - we know this from electrophysiology and fMRIs from humans - most fMRI experiments on visual processing begin by doing retinotopic mapping - when an animated stimulus sweeps over the visual field in a systematic way

Primary Visual Cortex (V1) - V1 is actually structured to represent orientation in a systematic way: it has a columnar structure where if an electrode is inserted vertically, the whole column of neurones has the same orientation preference. - But if the electrode is inserted, parallel to the cortical surface as the electrode is moved, it encounters neurones that have systematically different orientation preferences.

—> post-mortem V1 for an individual who has only one functioning eye, and the cortex is stained here so the areas that were recently active appear in black (so the black areas had inputs from a functioning eye) —>you can see that there are columns of neurones in V1 which have received inputs from each of the two eyes that are systematically interleaved.

There's also the concept of a hyper column in V1. - This is the area here indicated in Brown. - it's a chunk of V1, defined as representing all possible orientations for each of the two eyes. - then neighbouring hypercolumns would represent neighbouring areas on the retina.

Orientation Coding: - We know about columns and hyper columns from electrophysiology, but also from optical imaging. - Where in this case, a light is shone through a small hole in the skull to measure the changing colour of blood and tissue that's active, which allows a direct image of neural activity in response to the stimulation. - So different orientations of line are presented on a screen, and then the optical image reveals a systematic organisation of orientation preference for neurones and V1. - Data like this show the presence of what's called pinwheels, where orientation preferences change smoothly around a singularity, where cells with all different orientation preferences meet. - Orientation columns are so tiny, they're much smaller than a millimetre that you might think that it's impossible to find them in humans using fMRI (actually using very strong fMRI seven Tesla fMRI) - The same pinwheel structure has been found for neurone orientation preferences in human V1.

Orientation Channels

- Different neurones in V1 extract different orientations from the retinal image. - These images give you a sense of the information from the visual scene that's extracted by populations of V1 cells selected for different orientations. - Here are the original images and the second row preserves only the horizontal information on the third row, the vertical information. - This particular paper argues that for face perception, the most important information is carried by horizontally preferring neurones. - But the message is that different elements of visual information are all signalled by separate neural mechanisms.

Spatial Frequency Coding - Another key way the visual system breaks down visual information is by signalling different spatial frequencies separately, spatial frequency describes the spacing of light and dark contrasts in the visual scene. - So widely spaced contrast variation is low spatial frequency, while tightly spaced contrast variation is high spatial frequency. - The visual system is differently sensitive to different spatial frequencies, which you can see in this figure. - So for medium spatial frequencies, you can see the stripey pattern further up the diagram, and that means a low difference between the dark and li...