Complex cell

Last updated

Complex cells can be found in the primary visual cortex (V1), [1] the secondary visual cortex (V2), and Brodmann area 19 (V3). [2]

Contents

Like a simple cell, a complex cell will respond primarily to oriented edges and gratings, however it has a degree of spatial invariance. This means that its receptive field cannot be mapped into fixed excitatory and inhibitory zones. Rather, it will respond to patterns of light in a certain orientation within a large receptive field, regardless of the exact location. Some complex cells respond optimally only to movement in a certain direction.

These cells were discovered by Torsten Wiesel and David Hubel in the early 1960s. [1] They refrained from reporting on the complex cells in (Hubel 1959) because they did not feel that they understood them well enough at the time. [3] In Hubel and Wiesel (1962), [1] they reported that complex cells were intermixed with simple cells and when excitatory and inhibitory regions could be established, the summation and mutual antagonism properties didn't hold.

The difference between the receptive fields and the characteristics of simple and complex cells is the hierarchical convergent nature of visual processing. Complex cells receive inputs from a number of simple cells. Their receptive field is therefore a summation and integration of the receptive fields of many input simple cells, although some input is directly received from the LGN. [4] The manner through which simple cells are able to make up complex cells is not fully understood. A simple addition of receptive fields would result in complex cells manifesting observable, separate excitatory/inhibitory regions, which is not the case.

Discovery

The discovery of the complex cells in visual cortex began with experiments on a cat. Kuffler first shone small spots of light on a cat's retina. [5] With this, he was able to conclude that ganglion cells have concentric (active at high light levels) receptive fields. These cells also have either an on-center receptive field (excited when the stimulus is presented directly on the center of the receptive field) or off-center receptive field (excited when the stimulus is presented off the center of the receptive field). [5] Later, Hubel and Wiesel began their own experiments on cats to further solidify the knowledge on the visual receptive fields. One experiment recorded from anesthetized cats; these cats were paralyzed to stabilize their eyes. The cat then faced a screen where various patterns of white light were shone. Each cell's receptive fields were mapped for both eyes on sheets of paper. [6]

Other studies of complex cells have been performed by Movshon et al., [7] Emerson et al., [8] Touryan et al. [9] [10] and Rust et al. [11]

Simple vs. Complex Cells and Receptive Fields

With simple cells and simple receptive fields, the cells in visual cortex could respond in a way that can be noted from arrangements of excitatory and inhibitory regions in their receptive fields. What this means, essentially, is that the receptive fields are "simple" because there appears to be a relationship between the response of the cell and the receptive field mapped with small spots. Complex cells and complex receptive fields, on the other hand have a more complex response that does not exhibit that relationship. The results from the above experiment determined that simple fields have clear excitatory and inhibitory divisions, where light shone on an excitatory region increases the firing of a cell and light shone on an inhibitory region decreased firing of a cell. There is also evidence of summation properties, such as light shone across a larger region of either division resulted in a greater change in firing rate than light shone across a smaller region. It is also important to note that excitatory regions can inhibit inhibitory regions and vice versa, as well as it is possible to predict responses of the cells from a map of these areas.

On the contrary, complex cells and complex receptive fields are defined to be "not simple." These cell's response to a stimulus cannot be predicted as simple cells can, as they have no inhibitory and excitatory areas. Summation and the inhibition idea also do not often hold. For example, a horizontal slit was presented in the experiment, and it was found that a cell responded highly to this slit. On these complex cells, as long as the slit was horizontal, it did not matter where the slit was positioned on the receptive field. With simple cells, it would be expected that there would be a higher response to a wide slit. However, the opposite effect occurred: the firing of the cell actually decreased. It was also tested for orientation of the slit. For simple cells, it would be expected that as long as the slit covers the excitatory field, the orientation should not matter. Again, the opposite occurred where even slight tilts to the slit resulted in decreased response. [6]

Modeling Complex Cells

From various studies, including Movshon et al. in 1978 and even as early as the 1960s, simple cells have been able to be modeled with a linear model. [7] This would indicate that these simple cells undergo processes that calculate weighted sums of stimulus intensities where the weights are found from the receptive field. This stems from research by Enroth-Cugell & Robson in 1966 which modeled ganglion cells similar to P cells in primates (X cells) and ganglion cells similar to M cells in primates (Y cells). [12] Complex cells, on the other hand, are more complex and fall under a different model. Rather, it was noted that these cells perform nonlinear operations, which suggested that they have linear receptive fields, but instead sum a distorted output of subunits. It was found that complex cells shared similarities to Y cells, making this subunit model a promising candidate to model complex cells.

Movshon et al. in 1978 tested responses from simple cells to determine if the simple model for the X cells was a good fit. They later applied the same testing to complex cells, but used the Y cell (subunit) model instead. This model stated that each subunits could respond differently, but the converted responses would be offset in time, so it would sum to a constant value. It also stated that the response of the cells could not be predicted from the receptive field on its own. Complex cells appeared to match the subunit model, but still lacked the restriction that the receptive fields are linear. This was also tested by measuring the response of a cell when the stimulus contains two bars, which would help show the properties of the receptive field subunit. What they found was that by knowing these properties of the subunits, it was possible to predict spatial frequency selectivity, as was the case for simple cells. [13] Hence, complex cells could be modeled by the subunit model used for Y ganglion cells.

Other computational models of complex cells have been proposed by Adelson and Bergen, [14] Heeger, [15] Serre and Riesenhuber, [16] Einhäuser et al., [17] Kording et al., [18] Merolla and Boahen, [19] Berkes and Wiscott, [20] Carandini, [13] Hansard and Horaud [21] and Lindeberg. [22]

Related Research Articles

<span class="mw-page-title-main">Visual cortex</span> Region of the brain that processes visual information

The visual cortex of the brain is the area of the cerebral cortex that processes visual information. It is located in the occipital lobe. Sensory input originating from the eyes travels through the lateral geniculate nucleus in the thalamus and then reaches the visual cortex. The area of the visual cortex that receives the sensory input from the lateral geniculate nucleus is the primary visual cortex, also known as visual area 1 (V1), Brodmann area 17, or the striate cortex. The extrastriate areas consist of visual areas 2, 3, 4, and 5.

<span class="mw-page-title-main">Lateral geniculate nucleus</span> Component of the visual system in the brains thalamus

In neuroanatomy, the lateral geniculate nucleus is a structure in the thalamus and a key component of the mammalian visual pathway. It is a small, ovoid, ventral projection of the thalamus where the thalamus connects with the optic nerve. There are two LGNs, one on the left and another on the right side of the thalamus. In humans, both LGNs have six layers of neurons alternating with optic fibers.

<span class="mw-page-title-main">Torsten Wiesel</span> Swedish neuroscientist

Torsten Nils Wiesel is a Swedish neurophysiologist. With David H. Hubel, he received the 1981 Nobel Prize in Physiology or Medicine, for their discoveries concerning information processing in the visual system; the prize was shared with Roger W. Sperry for his independent research on the cerebral hemispheres.

<span class="mw-page-title-main">David H. Hubel</span> Canadian neurophysiologist

David Hunter Hubel was an American Canadian neurophysiologist noted for his studies of the structure and function of the visual cortex. He was co-recipient with Torsten Wiesel of the 1981 Nobel Prize in Physiology or Medicine, for their discoveries concerning information processing in the visual system. For much of his career, Hubel worked as the Professor of Neurobiology at Johns Hopkins University and Harvard Medical School. In 1978, Hubel and Wiesel were awarded the Louisa Gross Horwitz Prize from Columbia University. In 1983, Hubel received the Golden Plate Award of the American Academy of Achievement.

<span class="mw-page-title-main">Cortical column</span> Group of neurons in the cortex of the brain

A cortical column is a group of neurons forming a cylindrical structure through the cerebral cortex of the brain perpendicular to the cortical surface. The structure was first identified by Mountcastle in 1957. He later identified minicolumns as the basic units of the neocortex which were arranged into columns. Each contains the same types of neurons, connectivity, and firing properties. Columns are also called hypercolumn, macrocolumn, functional column or sometimes cortical module. Neurons within a minicolumn (microcolumn) encode similar features, whereas a hypercolumn "denotes a unit containing a full set of values for any given set of receptive field parameters". A cortical module is defined as either synonymous with a hypercolumn (Mountcastle) or as a tissue block of multiple overlapping hypercolumns.

The receptive field, or sensory space, is a delimited medium where some physiological stimuli can evoke a sensory neuronal response in specific organisms.

<span class="mw-page-title-main">Motion perception</span> Inferring the speed and direction of objects

Motion perception is the process of inferring the speed and direction of elements in a scene based on visual, vestibular and proprioceptive inputs. Although this process appears straightforward to most observers, it has proven to be a difficult problem from a computational perspective, and difficult to explain in terms of neural processing.

Neuronal tuning refers to the hypothesized property of brain cells by which they selectively represent a particular type of sensory, association, motor, or cognitive information. Some neuronal responses have been hypothesized to be optimally tuned to specific patterns through experience. Neuronal tuning can be strong and sharp, as observed in primary visual cortex, or weak and broad, as observed in neural ensembles. Single neurons are hypothesized to be simultaneously tuned to several modalities, such as visual, auditory, and olfactory. Neurons hypothesized to be tuned to different signals are often hypothesized to integrate information from the different sources. In computational models called neural networks, such integration is the major principle of operation. The best examples of neuronal tuning can be seen in the visual, auditory, olfactory, somatosensory, and memory systems, although due to the small number of stimuli tested the generality of neuronal tuning claims is still an open question.

The normalization model is an influential model of responses of neurons in primary visual cortex. David Heeger developed the model in the early 1990s, and later refined it together with Matteo Carandini and J. Anthony Movshon. The model involves a divisive stage. In the numerator is the output of the classical receptive field. In the denominator, a constant plus a measure of local stimulus contrast. Although the normalization model was initially developed to explain responses in the primary visual cortex, normalization is now thought to operate throughout the visual system, and in many other sensory modalities and brain regions, including the representation of odors in the olfactory bulb, the modulatory effects of visual attention, the encoding of value, and the integration of multisensory information. It has also been observed at subthreshold potentials in the hippocampus. Its presence in such a diversity of neural systems in multiple species, from invertebrates to mammals, suggests that normalization serves as a canonical neural computation. Divisive normalization reduces the redundancy in natural stimulus statistics and is sometimes viewed as an implementation of the efficient coding principle. Formally, divisive normalization is an information-maximizing code for stimuli following a multivariate Pareto distribution.

<span class="mw-page-title-main">Lateral inhibition</span> Capacity of an excited neuron to reduce activity of its neighbors

In neurobiology, lateral inhibition is the capacity of an excited neuron to reduce the activity of its neighbors. Lateral inhibition disables the spreading of action potentials from excited neurons to neighboring neurons in the lateral direction. This creates a contrast in stimulation that allows increased sensory perception. It is also referred to as lateral antagonism and occurs primarily in visual processes, but also in tactile, auditory, and even olfactory processing. Cells that utilize lateral inhibition appear primarily in the cerebral cortex and thalamus and make up lateral inhibitory networks (LINs). Artificial lateral inhibition has been incorporated into artificial sensory systems, such as vision chips, hearing systems, and optical mice. An often under-appreciated point is that although lateral inhibition is visualised in a spatial sense, it is also thought to exist in what is known as "lateral inhibition across abstract dimensions." This refers to lateral inhibition between neurons that are not adjacent in a spatial sense, but in terms of modality of stimulus. This phenomenon is thought to aid in colour discrimination.

The neocognitron is a hierarchical, multilayered artificial neural network proposed by Kunihiko Fukushima in 1979. It has been used for Japanese handwritten character recognition and other pattern recognition tasks, and served as the inspiration for convolutional neural networks.

<span class="mw-page-title-main">Simple cell</span> Beaker with Dilute Sulphuric Acid, Zinc and Copper Sheet is known as A Simple Cell

A simple cell in the primary visual cortex is a cell that responds primarily to oriented edges and gratings. These cells were discovered by Torsten Wiesel and David Hubel in the late 1950s.

<span class="mw-page-title-main">Hypercomplex cell</span>

A hypercomplex cell is a type of visual processing neuron in the mammalian cerebral cortex. Initially discovered by David Hubel and Torsten Wiesel in 1965, hypercomplex cells are defined by the property of end-stopping, which is a decrease in firing strength with increasingly larger stimuli. The sensitivity to stimulus length is accompanied by selectivity for the specific orientation, motion, and direction of stimuli. For example, a hypercomplex cell may only respond to a line at 45˚ that travels upward. Elongating the line would result in a proportionately weaker response. Ultimately, hypercomplex cells can provide a means for the brain to visually perceive corners and curves in the environment by identifying the ends of a given stimulus.

Feature detection is a process by which the nervous system sorts or filters complex natural stimuli in order to extract behaviorally relevant cues that have a high probability of being associated with important objects or organisms in their environment, as opposed to irrelevant background or noise.

<span class="mw-page-title-main">Parasol cell</span>

A parasol cell, sometimes called an M cell or M ganglion cell, is one type of retinal ganglion cell (RGC) located in the ganglion cell layer of the retina. These cells project to magnocellular cells in the lateral geniculate nucleus (LGN) as part of the magnocellular pathway in the visual system. They have large cell bodies as well as extensive branching dendrite networks and as such have large receptive fields. Relative to other RGCs, they have fast conduction velocities. While they do show clear center-surround antagonism, they receive no information about color. Parasol ganglion cells contribute information about the motion and depth of objects to the visual system.

<span class="mw-page-title-main">Orientation column</span>

Orientation columns are organized regions of neurons that are excited by visual line stimuli of varying angles. These columns are located in the primary visual cortex (V1) and span multiple cortical layers. The geometry of the orientation columns are arranged in slabs that are perpendicular to the surface of the primary visual cortex.

Joseph Anthony Movshon is an American neuroscientist. He has made contributions to the understanding of the brain mechanisms that represent the form and motion of objects, and the way these mechanisms contribute to perceptual judgments and visually guided movement. He is a founding co-editor of the Annual Review of Vision Science.

Binocular neurons are neurons in the visual system that assist in the creation of stereopsis from binocular disparity. They have been found in the primary visual cortex where the initial stage of binocular convergence begins. Binocular neurons receive inputs from both the right and left eyes and integrate the signals together to create a perception of depth.

Surround suppression is where the relative firing rate of a neuron may under certain conditions decrease when a particular stimulus is enlarged. It has been observed in electrophysiology studies of the brain and has been noted in many sensory neurons, most notably in the early visual system. Surround suppression is defined as a reduction in the activity of a neuron in response to a stimulus outside its classical receptive field.

Orientation selectivity is expressed by cells within the visual cortex, when such cells increase impulse or signal activity for specific oriented degree of shape presented within the visual field. Orientation selectivity can also be expressed by simple cells if the orientation of a stimulus is orthogonal to the preferred degree of orientation, which results in the inhibition of impulse activity.

References

  1. 1 2 3 Hubel DH, Wiesel TN (January 1962). "Receptive fields, binocular interaction and functional architecture in the cat's visual cortex". The Journal of Physiology. 160 (1): 106–54. doi:10.1113/jphysiol.1962.sp006837. PMC   1359523 . PMID   14449617.
  2. ↑ Hubel DH, Wiesel TN (March 1965). "Receptive Fields and Functional Architecture in Two Nonstriate Visual Areas (18 and 19) of the Cat". Journal of Neurophysiology. 28 (2): 229–89. doi:10.1152/jn.1965.28.2.229. PMID   14283058.
  3. ↑ Wiesel, David H.; Hubel, Torsten N. (2005). Brain and visual perception : the story of a 25-year collaboration ([Online-Ausg.]. ed.). New York, N.Y.: Oxford University Press. ISBN   978-0-19-517618-6.
  4. ↑ Palmer, Stephen E. (1999). Vision Science: Photons to Phenomenology. Cambridge: The MIT Press. p. 153. ISBN   978-0-262-16183-1.
  5. 1 2 Hubel DH, Wiesel TN (October 1959). "Receptive fields of single neurones in the cat's striate cortex". The Journal of Physiology. 148 (3): 574–91. doi:10.1113/jphysiol.1959.sp006308. PMC   1363130 . PMID   14403679.
  6. 1 2 Hubel DH, Wiesel TN (January 1962). "Receptive fields, binocular interaction and functional architecture in the cat's visual cortex". The Journal of Physiology. 160 (1): 106–54. doi:10.1113/jphysiol.1962.sp006837. PMC   1359523 . PMID   14449617.
  7. 1 2 Movshon JA, Thompson ID, Tolhurst DJ (October 1978). "Receptive field organization of complex cells in the cat's striate cortex". The Journal of Physiology. 283: 79–99. doi:10.1113/jphysiol.1978.sp012489. PMC   1282766 . PMID   722592.
  8. ↑ Emerson RC, Citron MC, Vaughn WJ, Klein SA (1987). "Nonlinear directionally selective subunits in complex cells of cat striate cortex". Journal of Neurophysiology. 58 (1): 33–65. doi:10.1152/jn.1987.58.1.33. PMID   3039079.
  9. ↑ Touryan J, Lau B, Dan Y (2002). "Isolation of relevant visual features from random stimuli for cortical complex cells". Journal of Neuroscience. 22 (24): 10811–10818. doi: 10.1523/jneurosci.22-24-10811.2002 . PMC   6758424 . PMID   12486174.
  10. ↑ Touryan J, Felsen G, Dan Y (2005). "Spatial structure of complex cell receptive fields measured with natural images". Neuron. 45 (5): 781–791. doi: 10.1016/j.neuron.2005.01.029 . PMID   15748852.
  11. ↑ Rust NC, Schwartz O, Movshon JA, Simoncelli EP (2005). "Spatiotemporal elements of macaque V1 receptive fields". Neuron. 46 (6): 945–956. doi: 10.1016/j.neuron.2005.05.021 . PMID   15953422.
  12. ↑ Enroth-Cugell C, Robson JG (December 1966). "The contrast sensitivity of retinal ganglion cells of the cat". The Journal of Physiology. 187 (3): 517–52. doi:10.1113/jphysiol.1966.sp008107. PMC   1395960 . PMID   16783910.
  13. 1 2 Carandini M (December 2006). "What simple and complex cells compute". The Journal of Physiology. 577 (Pt 2): 463–6. doi:10.1113/jphysiol.2006.118976. PMC   1890437 . PMID   16973710.
  14. ↑ Adelson E, Bergen J (1985). "Spatiotemporal energy models for the perception of motion". Journal of the Optical Society of America A. 2 (2): 284–299. Bibcode:1985JOSAA...2..284A. doi:10.1364/josaa.2.000284. PMID   3973762.
  15. ↑ Heeger DJ (1992). "Normalization of cell responses in cat striate cortex". Vis. Neurosci. 9 (2): 181–197. doi:10.1017/s0952523800009640. PMID   1504027. S2CID   22804285.
  16. ↑ Serre T, Riesenhuber M (2004). "Realistic modeling of simple and complex cell tuning in the HMAX model, and implications for invariant object recognition in cortex". Technical Report AI Memo 2004-017, MIT Computer Science and Artificial Intelligence Laboratory.
  17. ↑ Einhäuser W, Kayser C, König P, Körding KP (2004). "Learning the invariance properties of complex cells from their responses to natural stimuli". European Journal of Neuroscience. 15 (3): 475–486. doi: 10.1046/j.0953-816x.2001.01885.x . PMID   11876775. S2CID   7235110.
  18. ↑ Kording KP, Kayser C, Einhäuser W, Konig P (2004). "How are complex cell properties adapted to the statistics of natural stimuli?". Journal of Neurophysiology. 91 (1): 206–212. doi:10.1152/jn.00149.2003. hdl: 11858/00-001M-0000-0013-DA31-6 . PMID   12904330.
  19. ↑ Merolla P, Boahn K (2004). "A recurrent model of orientation maps with simple and complex cells". Advances in Neural Information Processing Systems (NIPS 2004): 995–1002.
  20. ↑ Berkes P, Wiskott L (2005). "Slow feature analysis yields a rich repertoire of complex cell properties". Journal of Vision. 5 (6): 579–602. doi: 10.1167/5.6.9 . PMID   16097870.
  21. ↑ Hansard M, Horaud R (2011). "A differential model of the complex cell". Neural Computation. 23 (9): 2324–2357. arXiv: 2012.09027 . doi:10.1162/neco_a_00163. PMID   21671791. S2CID   2075062.
  22. ↑ Lindeberg T (2020). "Provably scale-covariant continuous hierarchical networks based on scale-normalized differential expressions coupled in cascade". Journal of Mathematical Imaging and Vision. 62: 120–128. arXiv: 1905.13555 . doi: 10.1007/s10851-019-00915-x .
This page is based on this Wikipedia article
Text is available under the CC BY-SA 4.0 license; additional terms may apply.
Images, videos and audio are available under their respective licenses.