Overview of Receptive Fields
A wonderful thing about the art of painting is its explorations of the relationship between form and meaning. Flowers painted by Georgia O’Keefe, for example, are detailed studies of one isolated object, and are deeply resonant with sensuality and strength. Monet’s water lilies, on the other hand, show the relationship between flowers and their environment with tranquility and subtle grandeur. Our capacity for understanding both the object and the surrounding scene, the detail and the context, gives us a powerful ability to interact with the world in which we live. Neurons are able to achieve this in part because their receptive fields can distill both local and global features.
A “field” is a group of objects or properties that share a common characteristic. A field of crops, for example, could contain within it a field of corn and a field of wheat. Analogously, the receptive field of a neuron is the portion of the sensory field that affects the signalling of that neuron. For instance, the entire area that an eye can see is called the visual field, and the patch of the visual field that any single neuron monitors is that cell’s receptive field. In the auditory system, the range of all tones that we can hear makes up the auditory field, and the particular frequencies than a single neuron responds to make up that cell’s receptive field. In the somatosensory system, the whole surface of the body makes up the somatosensory field, and the receptive field of a single neuron is the patch of skin that it responds to.
Figure 1. Center-surround receptive fields arise from a pool of photoreceptors On-center and off-center fields in retinal bipolar and ganglion cells form by pooling the response of groups of photoreceptors. The photoreceptors can either act to excite (indicated in blue) or to inhibit (indicated in purple) a downstream bipolar cell. In an on-center bipolar cell, light hitting the central photoreceptors will be excitatory and light in the surround will be inhibitory. In an off-center bipolar cell, light in the center will be inhibitory, and light in the surround will be excitatory.
The shape of a neuron’s receptive field is defined by the synaptic inputs to that neuron. Connections that modify the passage of information between neighboring neurons are called local, or lateral connections. These lateral connections are often inhibitory, meaning that they cancel out part of the transmitted information. Inhibitory connections sculpt neural transmissions by cutting away parts of the underlying signal to sharpen aspects of the message. The receptive field of a neuron is initially a featureless patch, like an untouched block of marble. As the local network of neurons is connected, inhibitory signals begin to cut features out of the patch, like an artist cutting rough chunks out of the marble block. One “cut” might cause a neuron to ignore the color blue, or to ignore tones lower than a middle C. Increasing the complexity of the neural network can be compared to chipping away finer and finer flakes of marble; the remaining form is defined by the space that has been removed. A fine “cut” might allow the brain of a painter to distinguish the blue of the sky from the blue of a cornflower, or allow a musician to correct a middle C that is a few hertz too sharp.
Good examples of ways in which receptive fields are sculpted occur in the retina of the eye. The three layers of the retina work a bit like a company. Photoreceptor cells at the lowest level receive light from only a small patch in the visual field. That patch defines the photoreceptor’s receptive field. The photoreceptors report what they see to the next level of the retina, which is composed of bipolar, horizontal, and amacrine cells. Lateral connections from the horizontal and amacrine cells give the bipolar cells a receptive field shaped like two concentric discs. Each bipolar cell in the second layer is the manager of a group of photoreceptors, so these discs span the receptive fields of the photoreceptors underneath. As a manager, the bipolar cell can compare the reports of photoreceptors under its discs to detect spatial relationships between regions of light and dark. An on-center bipolar cell is strongly activated by a spot of light in the center disc of its receptive field surrounded by darkness in the outer disc. An off-center bipolar cell is strongly activated by light in the outer disc of its receptive field, with darkness in the center. These concentrically shaped receptive fields enable bipolar cells to detect edges, or transitions between regions of light and dark.
Figure 2. Center-surround receptive fields can detect edges As the gaze pans over a scene, edges will pass through the receptive field of center-surround cells in the retina. The cell will exhibit a characteristic activity pattern as the edge moves from the surround into the center and back out into the surround; that pattern of activity helps the brain identify the edge.
More complex receptive field shapes result from the combination of simpler fields. Connecting a series of concentric on-center fields, for example, yields a complex field that can detect a line. Complex receptive fields are a feature of cells that can detect the orientation of lines in primary visual cortex, and of cells in auditory cortex that respond to overtones and harmonics within a given tone. A general feature of cortical processing is that as we look further downstream in a pathway, receptive fields grow more complex and cells respond to increasingly higher order features of a stimulus.
Figure 3. Orientation sensitive cells in V1 combine upstream receptive fields The image of a horizontal line falling on the retina activates a group of photoreceptors arranged in a line. That activity is relayed to the lateral geniculate nucleus (LGN), where it activates a group of on-center cells also arranged in a line. In V1, a cortical cell combines the input from cells in the LGN, giving the cortical cell a linear, orientation specific receptive field.
The network architecture that gives rise to receptive fields provides the basis of our sensory perception. In the cortex, the architecture of each modality, or sensory system, includes a map that is a continuous representation of the corresponding sensory organ. In the auditory system, this is a tonotopic map of all of the frequencies that we can hear; in the somatosensory system, which is responsible for our sense of touch, it is a somatotopic map of the body; in the visual system, it is a retinotopic map of visual space as projected onto the retina. Complex receptive fields form when features from several of these cortical maps combine. In the superior colliculus, spatial maps from the auditory, visual, and somatosensory systems all intersect to provide information to cells that are responsible for orienting the head and eyes to sound.
The many ways in which receptive fields form are a testament to the brain’s ability to both dissect information by dividing a sensory field into small areas that can be monitored for detail, and to integrate information by combining features within and between receptive fields to detect the relationship between stimuli. It is this ability of the brain to understand both specific and general features which endows us with the flexibility to perceive so many facets of the world around us.