Receptive field

From Scholarpedia

Author: Dr. Jose-Manuel Alonso, SUNY State College of Optometry, New York, NY
Author: Dr. Yao Chen, SUNY State College of Optometry, New York, NY

The receptive field is a term originally coined by Hartline to define a restricted region of visual space where a luminous stimulus can drive electrical responses in a retinal ganglion cell. In Hartline’s own words, ‘Responses can be obtained in a given optic nerve fiber only upon illumination of a certain restricted region of the retina, termed the receptive field of the fiber’. After Hartline (1938), the term receptive field has been extended to other neurons in the visual pathway and other sensory pathways.

Visual receptive fields

Figure 1: Receptive field sizes of neurons in the primary visual cortex (V1) and inferotemporal cortex (IT) of a primate. The receptive fields are illustrated in dotted lines and the sizes are measured in visual degrees. When the viewing distance is 57 cm, one visual degree equals 1 cm at the point of fixation. Note that the receptive fields are much smaller in V1 neurons (0.5 – 2 degrees near the fovea) than IT neurons (~ 30 degrees).

The receptive field of a visual neuron is a two-dimensional region in visual space whose size can range from a few minutes of arc (a dot in this page at reading distance) to tens of degrees (the entire page). The receptive field size increases at successive processing stages in the visual pathway and, at each processing stage, it increases with the distance from the point of fixation (eccentricity).

Retinal ganglion cells located at the center of vision, in the fovea, have the smallest receptive fields and those located in the visual periphery have the largest receptive fields. The large receptive field size of neurons in the visual periphery explains the poor spatial resolution of our vision outside the point of fixation (other factors are photoreceptor density and optical aberrations). To become aware of the poor spatial resolution in our retinal periphery, try to read this line of text while fixating your eyes in a single letter. The letter that you are fixating is being projected at the center of your fovea where the receptive fields of retinal ganglion cells are smallest. The letters that surround the point of fixation are being projected in the peripheral retina. You will notice that you can identify just a few letters surrounding the point of fixation and that you need to move your eyes if you want to read the entire line of text.

Modern studies have expanded the term receptive field to include a temporal dimension. The spatiotemporal receptive field describes the relation between the spatial region of visual space where neuronal responses are evoked and the temporal course of the response. The relation between the spatial and temporal dimensions of the receptive field is particularly important to understand direction selective responses from neurons in primary visual cortex (Adelson & Bergen, 1985; Reid, Soodak, & Shapley, 1987; Watson & Ahumada, 1983).

Direction selective neurons respond to some directions of movement better than others. For example, a neuron may respond to a vertical line moving leftwards but not moving rightwards. The direction selective neurons generate visual responses with different time delays at different regions of the receptive field. Some regions respond faster to visual stimuli than others. As a consequence of these differences in response timing, a line moving from a slow to a fast region generates a stronger response than a line moving from a fast to a slow region. When the line moves in the optimal direction, the slow region, which is stimulated first, responds approximately at the same time asthe fast region, which is stimulated later. To make an analogy, imagine that two people are trying to say ‘response’ at the same time but one of them is speaking through a microphone that has a temporal delay of one second. The person that is using the delayed microphone has to say ‘response’ one second before the other for the two voices to fuse in unison. The result is a stronger ‘response’ than when the order is reversed.

Visual receptive fields are sometimes described as 3-dimensional volumes in visual space to include depth in addition to planar space. However, this use of ‘receptive field’ is less common and it is usually restricted to cortical neurons whose responses are modulated by visual depth.

Figure 2: On-center and Off-center receptive fields. The receptive fields of retinal ganglion cells and thalamic neurons are organized as two concentric circles with different contrast polarities. On-center neurons respond to the presentation of a light spot on a dark background and off-center neurons to the presentation of a dark spot on a light background.

Neurons at different stages in the visual pathway have receptive fields that differ not only in size but also in structure. The complexity of the receptive field structure, just as the receptive field size, increases at successive stages of the visual pathway. Most neurons in the retina and thalamus have small receptive fields that have a very basic organization, which resembles two concentric circles. This concentric receptive field structure is usually known as center-surround organization, a term that was originally coined by Kuffler (1953). On-center retinal ganglion cells respond to light spots surrounded by dark backgrounds like a star in a dark sky. Off-center retinal ganglion cells respond to dark spots surrounded by light backgrounds like a fly in a bright sky.

In primary visual cortex, receptive fields are much more diverse and more complicated than in the retina and thalamus. Only a few cortical receptive fields resemble the structure of thalamic receptive fields, while others have elongated subregions that respond to either dark or light spots, others respond similarly to light and dark spots through the entire receptive field and others do not respond to spots at all.

Hubel and Wiesel (1962) provided the first characterization of receptive fields in primary visual cortex and the first classification of cortical cells based on their receptive field structures. Some cortical cells respond to light and dark spots in different subregions of the receptive field and the arrangement of these subregions can be used to predict the responses of the cell to visual stimuli such as lines, bars or squared shapes. Cells with separate subregions that respond to either light or dark spots are called simple cells. All other cells in visual cortex that do not have separate subregions (the majority of the cells) are called complex cells.

Figure 3: Receptive fields of four primate V1 neurons (9^o-20^o eccentricity). The receptive field of each neuron was mapped with light spots (continuous lines, top panels) and dark spots (dotted lines, bottom panels). Unlike complex cells (c,d), simple cells (a.b) respond to light and dark spots in different regions of the receptive field (Figure taken from Anand et al., 2008).

Since Hubel and Wiesel (1962), other methods to classify cortical receptive fields have been proposed. However, to this date, no classification method has been widely adopted by the entire scientific community. Among all the classification methods after Hubel and Wiesel, the one that has been most widely used is based on the responses of cortical neurons to sinusoidal drifting gratings. Some cortical neurons respond to the sinusoidal changes in luminance by generating a rectified sinusoidal response (which is a rough linear replica of the stimulus) while others respond by increasing the mean firing rate. A quantitative measurement of response linearity can be obtained by Fourier analysis. Response linearity is bimodally distributed (Skottun et al., 1991).

The great diversity of receptive fields in primary visual cortex makes it difficult to correlate neuronal classes with receptive field properties, as is currently possible in the retina (e.g. Masland, 2001). Neurons in primary visual cortex can respond selectively to different attributes of the visual scene such as line orientation, direction of movement, luminance contrast, stimulus velocity, color, retinal disparity and spatial frequency (frequency of black and white stripes in a degree of visual space).

Figure 4: Linear and non-linear V1 neurons in primate. The visual responses of linear neurons (top row) resemble a rectified replica of the sinusoidal stimulus (drifting grating). In contrast, the visual responses of nonlinear neurons (bottom row) resemble a step function. Left. Cartoon illustrating the changes in the amplitude of the stimulus (continuous lines) and response (dotted lines) across time. Middle and right. Raster plots (top panels) and peri-stimulus time histograms (PSTHs, bottom panels) for the four same cells illustrated in Figure 3 (Fig. 3a: top-middle, Fig 3b: top-right, Fig 3c: bottom-middle, Fig 3d: bottom-right). Each tick in the raster plot represents a spike. Spont: spontaneous activity (spikes generated in the absence of a sinusoidal modulation of the stimulus).

Most neurons in primary visual cortex respond to moving lines and are selective to line orientation. Some neurons are sharply tuned to orientation and fail to respond to lines that are just a bit tilted from their preferred orientation while other cortical neurons are broadly tuned and respond to a broad range of orientations. The selectivity of each neuron to line orientation and other parameters are determined to a great extent by the receptive field structure. A very active area of research aims to build realistic models of receptive field structures that can explain neuronal responses to different stimuli. The most successful models to date were built for neurons at the earliest stages of the visual pathway. For example, the receptive fields of retinal and thalamic neurons can be modeled quite accurately with a difference of Gaussians (DOG, Rodieck, 1965). Also, the receptive fields of visual cortical neurons that receive direct input from the thalamus can be modeled with Gabor functions (Jones & Palmer, 1987).

Figure 5: Orientation tuning in V1 neurons. Polar plots of two neurons with sharp (a) and broad (b) orientation tuning measured with drifting sinusoidal gratings. The radial coordinate illustrates firing rate and the angle the direction of movement. The PSTHs show neuronal responses to gratings drifting for one second in four different directions of movement. Scale bars refer to the radius of each polar plot.

The term receptive field becomes increasingly difficult to apply as information progresses through the visual pathway. Neurons in higher cortical areas respond to stimuli presented over large regions of the visual field and can be more selective to the identity of the stimulus than to its physical location. For example, neurons in the inferotemporal cortex respond selectively to objects and faces (Bruce, Desimone, & Gross, 1981; Desimone, Albright, Gross, & Bruce, 1984; Tsao & Livingstone, 2008). A remarkable example of selectivity for stimulus identity was recently reported for a neuron in temporal cortex. This neuron responded selectively to the identity ‘Halle Berry’, either presented as a face or simply as the written name of the actress (Quiroga, Reddy, Kreiman, Koch, & Fried, 2005).

Somatosensory receptive fields

The receptive fields of somatosensory neurons share much in common with the receptive fields of visual neurons. As for visual neurons, the somatosensory receptive fields are defined as a restricted 2-dimensional region of space where a stimulus can evoke a neuronal response. In somatosensory neurons, however, space refers to a region of the body and the stimulus can be touch, vibration, temperature or pain.

Similar to visual neurons, the receptive fields of somatosensory neurons are smaller in the regions of the body where the perceptual spatial resolution is highest. The fingertips have the highest spatial resolution (and the smallest receptive fields) while the thigh and calf region have the lowest spatial resolution (and largest receptive fields). The spatial resolution to light-touch stimulation can be evaluated by measuring two-point discrimination thresholds. The subject has to report whether the skin is touched either with one or two pointy objects that are closely spaced. When the distance between the two objects is small, it is not possible to reliably distinguish between one or two objects touching. The minimum distance that is required to distinguish two pointy objects is called the two point discrimination threshold. The two point discrimination threshold is less than 5 mm at the finger tips and is about 40 mm at the thigh.

As in the visual system, the receptive fields in the somatosensory thalamus have center-surround organization and those in the somatosensory cortex have more complex receptive field structures that make the neurons selective to the orientation and direction of motion of a stimulus.

Auditory receptive fields

In auditory physiology, the term receptive field is frequently used with two different meanings. As a first meaning, an auditory receptive field can be defined as the region in auditory space where a stimulus is able to evoke a response in an auditory neuron (auditory spatial receptive field). As the second meaning, an auditory receptive field can also refer to the range of sound frequencies that most optimally stimulate the neuron (auditory spectrotemporal receptive field).

Auditory spatial receptive fields resemble visual and somatosensory receptive fields in that they represent an area of auditory space where a sound generates a neuronal response. Like with visual and somatosensory receptive fields, spatial receptive fields at early stages in the auditory pathway have center-surround organization. For example, some auditory neurons in the midbrain respond to sounds presented at a defined region of auditory space, which is the receptive field center, and the response is reduced when the stimulus is presented in a region surrounding the center, which is the receptive field surround. The center-surround receptive fields of auditory neurons cover a much larger region of space than visual and somatosensory receptive fields with similar center-surround organization. Auditory spatial receptive fields tend to be located in front of the animal and they can be restricted to a single quadrant in the contralateral side of the brain where the neuron is recorded (Knudsen and Konishi 1978).

Auditory receptive fields can also refer to the frequency range of sounds that most optimally stimulate the neuron (spectrotemporal receptive field). The use of spectrotemporal receptive fields reflects a conspicuous difference between the auditory and visual/somatosensory systems. Unlike in the visual and somatosensory systems, in the auditory system space is not topographically mapped in the sensory organ, the cochlea. While the retina and the skin have a precise representation of visual and body space, neurons in the cochlea have a precise representation of sound frequency. The representation of sound frequency is organized like a piano scale: lower tones are represented at the apex of the cochlea and higher tones at the base. Neurons at different stages of the auditory pathway can be very sensitive to small variations in sound frequency and their responses can have different time courses. Both the frequency range and time course of the response can be quantitatively represented in a map of the spectrotemporal receptive field.

Note for the readers

This is our first draft of the page receptive field. We would like to hear from you to help us fill the missing gaps (including scientific references).

Glossary

Auditory spatial receptive field: The region of space where a sound can generate a response in an auditory neuron.

Auditory spectrotemporal receptive field: Spectrum of sound frequencies that generate a response in an auditory neuron (represented as a function of the time-course of the response).

Broadly tuned: Refers to neurons that respond similarly to wide range of variations within a given stimulus dimension. For example, neurons that have broad orientation tuning respond similarly to all line orientations (Fig. 5b). Neurons that have broad spatial frequency tuning respond similarly to a wide range of spatial frequencies.

Cochlea: A portion of the inner ear which is a spiraled, hollow, conical chamber of bone. The auditory sensory neurons are located inside the cochlea.

Complex cells: Neurons in the primary visual cortex that cannot be classified as simple cells (they do not respond to light and dark spots in different regions of the receptive field).

Directional selective: Neurons that respond strongly to a specific direction of movement and fail to respond (or respond weaker) to the opposite direction.

DOG: A function that results from the difference of two Gaussian functions. DOG stands for difference of Gaussians.

Eccentricity: The distance between the receptive field center of a given neuron and the center of vision (point of fixation or fovea).

Electrical responses: Electrical activity that neurons generate in response to a sensory stimulus. The term ‘electrical response’ is usually reserved for membrane depolarizations that lead to action potentials (also called spikes) in an individual neuron. Such spikes can be extracellularlly recorded with microelectrodes, which is the technique most frequently used to map neuronal receptive fields.

Fourier analysis: A mathematical method to characterize general functions by sums of simpler circular functions (sinusoidal and cosinusoidal functions). It is used to extract specific frequency components from the PSTHs of the neuronal responses and measure the amplitude and phase of each component.

Fovea: A small region of the retina (~1 mm diameter for the human fovea) where cone photoreceptors are most densely packed to provide the highest visual acuity.

Gabor functions: Functions that result from the multiplication of a sinusoidal function with a Gaussian function.

Mean firing rate: Total number of spikes averaged over time and over multiple stimulus presentations. It is usually measured in spikes per second.

Minute of arc: One sixtieth of a visual degree.

Receptive field: A specific region of sensory space in which an appropriate stimulus can drive an electrical response in a sensory neuron.

Rectified sinusoid: A sinusoidal function that is half-wave rectified (all negative values are set to zero).

Retinal ganglion cell: Neuron located in the inner part of the retina (part facing the pupil), which carries visual information from the eye to the deep structures of the brain.

Sharply tuned: Refers to neurons that respond only to a narrow range of variations in a given stimulus dimension. For example, neurons that have sharp orientation tuning respond only to a narrow range of line orientations (Fig. 5a). Neurons that have sharp spatial frequency tuning respond only to a narrow range of spatial frequencies.

Simple cells: Neurons in the primary visual cortex, whose receptive fields have separate subregions that respond either to light or dark spots. The discovery of simple cells in visual cortex and the first use of the term ‘simple cell’ date back to Hubel and Wiesel (1962). Hubel and Wiesel defined simple cells as cells in the cat primary visual cortex whose receptive fields meet four different criteria: 1) the receptive fields can be subdivided into distinct excitatory and inhibitory regions; 2) there is summation within the separate excitatory and inhibitory parts; 3) there is antagonism between excitatory and inhibitory regions; 4) it is possible to predict the responses to stationary or moving spots of various shapes from a map of the excitatory and inhibitory areas.

Spatial frequency: Frequency of black and white stripes in a degree of visual space. It is measured in cycles per degree. One cycle is a set of a black and a white stripe. For example, a pattern of black-white-black-white stripes contained in a degree has a spatial frequency of 2 cycles per degree.

Spatiotemporal receptive field: A spatial receptive field plotted at different time delays between stimulus and neuronal response.

Visual degree: The amount of visual space covered by a cone of 1 degree angle with its apex located at the fovea of the retina. One visual degree covers a circle of 1 cm diameter when the distance between the eye and the fixation point is 57 cm. When the distance is 114 cm distance, one visual degree covers a circle of 2 cm diameter.

Visual periphery: Part of the visual field that is projected on the non-foveal retina (also called peripheral retina).

References

• Adelson, E H and Bergen, J R (1985). Spatiotemporal energy models for the perception of motion. J Opt Soc Am A 2(2): 284-299.
• Anand, S et al. (2008). Relating spontaneous firing rate to response linearity and stimulus selectivity in the awake primary visual cortex. Journal of Vision (Submitted): PAGES.
• Bruce, C; Desimone, R and Gross, C G (1981). Visual properties of neurons in a polysensory area in superior temporal sulcus of the macaque. J Neurophysiol 46(2): 369-384.
• Desimone, R; Albright, T D; Gross, C G and Bruce, C (1984). Stimulus-selective properties of inferior temporal neurons in the macaque. J Neurosci 4(8): 2051-2062.
• Hartline, H K (1938). The response of single optic nerve fibers of the vertebrate eye to illumination of the retina. Am J Physiol 121: 400-415.
• Hubel, D H and Wiesel, T N (1962). Receptive fields, binocular interaction and functional architecture in the cat's visual cortex. J Physiol 160: 106-154.
• Jones, J P and Palmer, L A (1987). An evaluation of the two-dimensional Gabor filter model of simple receptive fields in cat striate cortex. J Neurophysiol 58(6): 1233-1258.
• Knudsen, E I and Konishi, M (1978). Center-surround organization of auditory receptive fields in the owl. Science 202(4369): 778-780.
• Kuffler, S W (1953). Discharge patterns and functional organization of mammalian retina. J Neurophysiol 16(1): 37-68.
• Masland, R H (2001). Neuronal diversity in the retina. Curr Opin Neurobiol 11(4): 431-436.
• Quiroga, R Q; Reddy, L; Kreiman, G; Koch, C and Fried, I (2005). Invariant visual representation by single neurons in the human brain. Nature 435(7045): 1102-1107.
• Reid, R C; Soodak, R E and Shapley, R M (1987). Linear mechanisms of directional selectivity in simple cells of cat striate cortex. Proc Natl Acad Sci U S A 84(23): 8740-8744.
• Rodieck, R W (1965). Quantitative analysis of cat retinal ganglion cell response to visual stimuli. Vision Res 5(11): 583-601.
• Skottun, B C et al. (1991). Classifying simple and complex cells on the basis of response modulation. Vision Res 31(7-8): 1079-1086.
• Tsao, D Y and Livingstone, M S (2008). Mechanisms of face perception. Annu Rev Neurosci 31: 411-437.
• Watson, A B and Bocca, Ahumada (1983). A look at motion in the frequency domain. NASA Technical Memorandum, 84352.

Further reading

None.

External links

http://en.wikipedia.org/wiki/Receptive_field

See also

None.