Processing Visual Stimuli in the Brain Glenn Mason-Riseborough (17/4/1997) One day, while walking down Queen Street, you are delighted to see an old friend whom you lost contact with many years ago. There are many parts of the brain that are involved in processing the physical stimulus of the above scenario into a perception of a friend’s face. To understand the processes involved, a number of different levels of processing need to be discussed. Firstly, there is the processing that takes place inside an individual neuron, and between single neurons. This involves concepts such as resting membrane potential, postsynaptic potentials, action potentials and synaptic transmission. Another level of processing occurs inside separate structures of the visual system. This includes such structures as the retina, thalamus, and visual cortex. We will also look at how these separate structures are connected together to give us a perception of our friend’s face. The main task of this essay is to describe the pathways involved in processing the information. Due to space constraints, this essay will give a very simplified account of processes inside the neuron. The remainder of the essay will discuss the processing involved from light entering the eye to the visual cortex. Figure 1: Schematic diagram of typical miltipolar neuron. The resting membrane potential is the voltage difference between the inside and outside of a neuron. Figure 1 shows the structure of a typical neuron. This difference is about -70 mV, and at this voltage difference, the neuron is said to be polarised. This voltage difference is detected by inserting an electrode inside the neuron and another electrode just outside that neuron. Connecting these electrodes to a voltage detector such as an oscilloscope or multimeter gives the neuron a voltage difference of -70 mV with respect to outside the neuron (Pinel, 1993, p 99). There are four ions that contribute to this polarised state, these ions are anions (A-) (negatively charged protein ions that are manufactured inside the neuron), chloride (Cl-), sodium (Na+), and potassium (K+) (Pinel, 1993, p 100). These ions are acted on by four separate forces. These forces are random motion, electrostatic pressure, differential permeability, and a sodium- potassium pump (Pinel, 1993, p 100). Random motion refers to the fact that the ions are in random motion, and with time they are likely to become distributed evenly throughout the medium. The second force, electrostatic pressure, is the force with which like charges repel. This means that sodium and potassium ions repel each other and thus become evenly distributed. The same rules apply for chloride and anions. Differential permeability refers to the relative difficulty or ease in which the ions can pass through the cell membrane. Potassium and chloride pass through readily, sodium passes with difficulty and anions do not pass through at all. The above three forces are all passive forces, that is, they do not require energy. The sodium-potassium pump requires energy to operate. This pump moves sodium out of the neuron and potassium inside the neuron. A number of studies (for example Hodgkin and Keynes, 1955, cited in Pinel, 1993, p 101) have confirmed the existence of this pump. Figure 2: Schematic diagram of a synapse. Postsynaptic potentials are created when a neuron absorbs neurotransmitters that have been released from other neurons at a synapse. These neurotransmitters can either decrease the resting membrane potential (depolarisation) or increase the resting membrane potential (hyperpolarisation). Depolarisations increase the chance of the neuron firing; these are called excitatory postsynaptic potentials. Hyperpolarisations decrease the chance of firing; they are inhibitory postsynaptic potentials. The excitatory and inhibitory postsynaptic potentials resulting from different synapses are summed together. If the resting membrane potential is decreased beyond a certain threshold (approximately -65 mV (Pinel, 1993, p 105)) at the axon hillock, then an action potential occurs. When an action potential occurs, voltage-gated sodium channels open and sodium rushes into the neuron. The membrane potential quickly changes to +50 mV, and the potassium channels are opened. Potassium is driven out of the neuron and some chloride enters because of the net positive voltage difference. The sodium channels are then closed, and the neuron is repolarised by the potassium ions. The neuron becomes hyperpolarised before the potassium channels finally close. Polarisation is quickly re-established by random motion. This process is repeated along the length of the axon until it reaches the terminal button. Myelination of axons increases the speed and distance of transmission by only allowing the sodium and potassium channels to open at the (unmyelinated) nodes of Ranvier. Signals are transmitted from one neuron to the next via chemical neurotransmitters at the synapse (figure 2 shows the major structures of a synapse). When an action potential arrives at the terminal button of the sending neuron, synaptic vesicles containing neurotransmitter attach to the presynaptic membrane and the neurotransmitters are released into the synaptic cleft. The neurotransmitter then binds onto receptor molecules on the surface of the postsynaptic membrane. How this is accomplished depends on the type of neurotransmitter. The end result is that the neurotransmitter produces either a depolarisation or a hyperpolarisation of the receiving neuron. Figure 3: Drawing of a cross section of a human eye. When you are walking down Queen Street, light is constantly entering your eyes through the pupil, refracting through the lens to focus on the retina (figure 3 is a drawing of a sectional view of a human eye). The retina consists of five layers of cells. These cells are (from back of retina to front) receptors, horizontal cells, bipolar cells, amacrine cells, and retinal ganglion cells (see figure 4 for a diagram of the retinal cells). When light gets to the retina it must pass through the four other layers of cells before it gets to the receptors. This results in a very slight distortion in the light that the receptors receive. The receptors absorb the light and convert it into the electrical and chemical system (described in the preceding paragraphs) that the brain uses for signal transmission. There are two types of receptor cells -- rods and cones. This distinction refers to the shape of the cells. Rods operate optimally in dim light (scotopic system) and cones in brighter light (photopic system). A difference that may account for this, is that hundreds of rods connect to a single ganglion cell, but only a few cones may connect to each ganglion cell (Pinel, 1993, p 195). Support for this is from case studies of patients who lack either rods or cones (Pinel, 1993, p 194). Thus we know that the scotopic system does not provide the detail and colour of the photopic system, and if we had been walking down Queen Street at night we may not have recognised our friend. Another difference between rods and cones is that cones are primarily in the fovea and rods are located only in the periphery. Thus it is a great deal more likely that we would recognise our friend if we are looking directly at him/her. Colour is an important aspect in recognition of our friend. For example, do they have red, brown or blonde hair? In 1802 Young (cited in Pinel, 1993, p 216) proposed the trichromatic theory which was confirmed by a number of researchers in the early 1960s (Pinel, 1993, p 216). This theory states that there are three different types of cone receptors, each of which responds to different frequencies. The rich variety of colours we see is achieved by combining the signals from these receptors. Figure 4: Diagram of the cellular structure of the retina. When the receptors are stimulated by light photons they send an inhibitory signal to the bipolar cells. This decreases the chance that the bipolar cells send an inhibitory signal to the ganglion cells which consequently fire. Horizontal cells and amacrine cells provide contrast, thus emphasising the contours and areas of light and dark of our friend’s face. This contrast is achieved by being connected horizontally across the retina and inhibiting the neighbours of a photoreceptor that fires. The axons of the ganglion cells form the optic nerve which leaves the retina at the blind spot and terminates at the thalamus. The blind spot is an area on the retina which contains no photoreceptors; the brain is forced to infer this spot using information from the surrounding area. Hence, if we were only looking through one eye, and our friend was standing in exactly the right spot, it may seem as if our friend had no head. The thalamus is simply a relay station for the incoming signals. The axons of the retinal ganglion cells terminate in the lateral geniculate bodies of the thalamus. The right lateral geniculate takes care of the signals from the left visual field and vice versa. This is because half the axons cross at the optic chiasm (figure 5 shows the retinal- geniculate-striate system). The axons of the retinal ganglion cells synapse at two different areas of the thalamus (Johnson, 1997). ‘P’ type axons are those that carry colour and detail and mainly originate from cone receptor cells. These terminate onto the parvocellular level of the lateral geniculate. ‘M’ type axons carry movement information and are mainly from rods. These terminate on the magnocellular level of the lateral geniculate. Neurons from these areas of the thalamus project axons that terminate in the lower part of cortical layer IV of the primary visual cortex. Signals from ‘P’ type axons would be mainly used in the perception of our friend’s face. The retina-geniculate-striate is not the only visual pathway in the brain. Other pathways studied include one which travels from the retina, through the superior colliculi of the midbrain and pulvinar nuclei of the thalamus, to the secondary visual cortex (Pinel, 1993, p 228). Figure 5: Diagram of the retina-geniculate-striate pathway. Cells from the retina to lower layer IV of the primary visual cortex all have similar receptive fields. These receptive fields are round with separate central and peripheral parts. The cell either has an on response to the centre and an off response to the periphery, or an off response to the centre and an on response to the periphery (Hubel and Weisel, 1979, cited in Pinel, 1993, p 206). This suggests that the function of many of the neurons in these areas is to provide contrast. This connects with what was discussed above in relation to horizontal and amacrine cells. Higher level processing cells in the visual cortex have receptive fields which consist of straight lines rather than concentric circles (Pinel, 1993, p 207). These cells fire due to stimulus of lines at different orientations and movements, thus different cells will respond to different edges of our friend’s face, and different faces will elicit slightly different cell firings. Convention dictates that sensation is subcortical and perception is cortical (Pinel, 1993, p 224). Thus, up until now this essay has primarily discussed how the visual image of our friend is detected. It is only in the visual cortex that this image is interpreted, and recognition of our friend takes place. It is not at all clear cut how this interpretation and recognition actually take place. The theories that are developed can only be inferred from observations of humans and animals and by comparing ‘healthy’ subjects to subjects with reported recognition problems or lesions. Figure 6: Diagram of the visual areas of the human neocortex. From the primary visual cortex, visual signals seem to split and travel along two major routes (Ungerleider and Mishkin, 1982, cited in Pinel, 1993, p 229). Both of these routes travel via the prestriate cortex, with one leading to the inferotemporal cortex, and the other to the posterior parietal cortex (Pinel, 1993, p 229). Figure 6 is a diagram of a human brain showing the routes these signals take. Prestriate cortex and inferotemporal cortex are both considered to be secondary visual cortex, while posterior parietal cortex is considered to be association cortex because it also receives input from auditory cortex and somatosensory cortex (Pinel, 1993, p 229). The cortical visual pathways are divided into five areas designated V1 through to V5 (Tippett, 1997). V1 is the primary visual cortex, V2 to V5 are areas in the prestriate cortex. V2 is concerned with motion, orientation and colour, V3 with orientation, V4 with colour and V5 with motion. The route to inferotemporal cortex seems mainly to be connected to object recognition (V1, V2, V3, V4). The route to posterior parietal cortex seems to be connected to perception of movement and spatial location (V1, V2, V5) (Desimone and Ungerleider, 1989, cited in Pinel, 1993, p 229). As we walk towards our friend the later route is crucial for us to orientate towards them, but it is the former route which produces the recognition. There are some interesting case studies of people with recognition deficits due to lesions in certain areas as described above. These cases are in no way a black and white issue of a certain lesion producing a certain recognition deficit, however general rules can be applied. For example, prosopagnosics (people who can not recognise familiar faces) tend to have bilateral damage to the inferior prestriate region and also to adjoining areas of the inferotemporal area (Damasio, 1985 and Meadows, 1974, cited in Pinel, 1993, p 232). Other studies for example by Kendrick and Baldwin in 1987 (Pinel, 1993, p 232) show neurons in these areas in sheep respond to faces. All these cases seem to suggest that it is the prestriate cortex and inferotemporal cortex which do the processing for us to perceive our friend’s face. The seemingly instantaneous process of seeing our friend and recognising him or her is actually the result of a lot of processing in many parts of our brain. The light that reflects off our friend’s face and hits our retina is broken down into separate parts. Each part is processed in different regions before being reconstructed into an image that we recognise. Much research has gone into finding these regions and discovering how they work. Results indicate that the major visual pathway is the retinal-geniculate-striate pathway, however other pathways are also involved. Damage in any part of this main pathway could result in us not seeing parts or all of our friend. Even if we do see them, there is no guarantee that we will recognise what we see. Studies of patients with damage to areas in the prestriate cortex or inferotemporal cortex report them not being able to recognise familiar faces. The patients even go as far as describing in detail the features of familiar faces, but then still not able to say who the face belongs to. What we are left with is a picture of visual processing that seems counter-intuitive. No longer is vision an integrated all or nothing process, but a complex hierarchy of many different interconnected processes. References: Johnson, B. (1997). Unpublished lecture notes for University of Auckland paper 461.230. Pinel, J. P. J. (1993). Biopsychology (2nd edition). Allyn and Bacon. Tippett, L. J. (1997). Unpublished lecture notes for University of Auckland paper 461.230.