Peirce, Neurons, and Warhol’s soup cans

Brian Coyle
3 min readNov 9, 2021

Symbol, index, and icon was a triad created by Charles Sanders Peirce. It is a basis of semiotics, the study of signs and their meanings. Symbols do not resemble their object, but refer to it by convention. The alphabet is an example. Whatever their origin, letters no longer resemble the kinds of sounds they generate. Indicies are a distilled version of the object, which include some aspect that indirectly resembles it. A index of genetic code uses letters, but they are structured in the same way as DNA elements in a chromosome. A road map does the same thing. Icons include aspects that directly resemble the object they stand for. The iconic images on US and many other coins and bills are streamlined, simplified portraits of real people. The coin itself is a symbol, referring to abstract value.

Memory is held in neural structures of various kinds. The animal brain uses stable ensembles of many neurons as memory substrates, especially for long-term memory. These neocortical networks are updated continuously as an animal learns. The nature of neuronal populations suggests their dynamic structure contains information. This process involves abstraction, the coding of stimuli in patterns the brain understands. It maps to Peirce’s symbol level, that involve cues whose meaning emerges in a network or pattern of cues.

While many elements of neural networks are in flux, with dendrictic spines being formed and eliminated, a signficant number do not participate in this churning. These synapses persist over an animal’s lifetime, more common in the neocortex than hippocampus. Long-term memories may be enabled by stable spines and neurons that store basic features, while details are represented in elements that are more plastic and changeable. This has actually been called the “memory indexing” hypothesis, in which neocortical storage sites can remake neocortical activity patterns to retrieve memory. Like the Peircian index, its a distilled version than can be reconstituted. Using a map, one can travel a path known from its representation. With a stable dendritic backbone, an animal can perform well in navigation tasks despite changing neural activity in motor areas.

In studies of sensory processing, researchers debate two processes. Is vision dominated by the activity of large neuronal ensembles in which a large fraction of component neurons fire, or are percepts identified with a small neuronal ensemble, whose member neurons respond to specific features, objects or concepts? Note that again the first process involves neurons tuned to the group pattern, not the object being sensed. This is the symbol level, while the second process involves elements that represent actual features, like an index.

Finally, a third level is proposed, sometimes called “grandmother cells.” Though controversial, these are neurons that respond to one object, one person, perhaps independently of the angle of gaze. Widely dismissed a decade ago, recent research found temporal pole region neurons behave like memory cells that respond to familiar faces as rapidly as sensory cells respond to visual stimuli. These parallel the icon level of Peirce’s semiotics.

Artists are privaledged to work with different levels of perception, and by extension interact with different ways the brain processes sensory information. Like philosophers such as Peirce, they may intuit insights that reveal how our minds work. Consider the artist Andy Warhol, whose paintings represent iconic imagery of late 20th century consumer culture. By presenting images that triggered iconic, immediate response, they proved compelling. But their layered, abstracted aspect pulled them into a zone of reflection. While the iconic soup can was immediately detected by neurons habituated to seeing it in advertising and shopping, it’s curious composition led to a permentant, more complex memory.

Today, a group of Campbell’s soup cans is processed as an index of American culture, not just an icon for dinner. It has not become a symbol, whose meaning is completely abstracted, held in activity patterns that generate higher level meanings of their own. Perhaps it will travel this route. Chinese characters are almost entirely abstract, but evidence points to at least some having iconic origin a thousand years ago. One day soup cans could become part of a future computer language, a symbol that might mean withdrawing something from storage over and over.

The source of icons in real life provides the first input, then a stable network forms around an indexical version, and maybe finally the network itself becomes what matters, whatever stimulus remains referring to it as a symbol.

--

--