Navigation has typically involved something more technical than biological, especially in relation to traversing and remembering spaces. From compass and map to astrolabe and the Global Positioning System (GPS), humans have long relied on a variety of devices to get themselves or their projectiles from here to there. But these tools are not the only game in town—biological navigation, is crucial for the everyday life, movement, and survival of a myriad of species, not just human. Nowadays, this interplay between technical and biological navigation is increasingly blurry. What are the characteristics of navigation that we encounter along the gradient between the technical and the biological, between positioning and memory? To answer this, it helps to put the discourse of “cognitive mapping” into dialogue with advances in neuroscience and artificial intelligence, where scientists now speak of “an inner GPS.” GPS, in this sense, is a misaligned metaphor for a cognitive map, a figure that has long been operational within the fields of architecture, urbanism, and human-computer interaction. However, this conflation warrants revisiting and critiquing their fundamental concepts once again.
We are living in a moment when artificial intelligence—the technical simulation of a biological brain—threatens to absorb and replace many fields, disciplines, and control systems. Simultaneously, the same technologies and algorithms that aim to simulate biological brains (and want to exceed their capabilities) still can’t map the brain of many species, including human. Researchers claim the human brain is the most complex organ in any living creature—a network of trillions of brain cells or neurons housed in a jello-like framework. But networks are only one way of describing the brain. It is therefore important to interrogate the relationships between the elements of this binary—biology and technology—that build these networks, and their cognitive capacities.
From Inner Beauty to Inner GPS
There is a grand tradition of making art out of human anatomy, from the comic grotesqueries of Vesalius to the exquisite line drawings of Cajal. The twenty-first century is no exception. Just because these images depend on expensive machines, doesn’t make the scientist a passive observer, or one that no longer thinks about aesthetics.1
—Carl Schoonover
While artistic and scientific paradigms are hugely important in the history of imaging the brain, so are the metaphors used in describing this particular part of human anatomy. Consider the press releases announcing two Nobel Prizes given in “Physiology or Medicine.” In 1906, this prize was awarded to Camillo Golgi and Santiago Ramón y Cajal “for revealing the inner beauty of the nervous system. By developing methods that could color and highlight its key components, Golgi and Cajal allowed the anatomy of the nervous system to be observed and documented in precise detail.”2 This refers to their drawings, which Golgi made from staining the physical brain and which Cajal made from memory after looking at neurons in a microscope. The sheer beauty of the brain they drew is hard to contest.
108 years later, in 2014, the Nobel press release was more prosaic in describing the achievements that led to the same prize: “How do we know where we are? How can we find the way from one place to another? And how can we store this information in such a way that we can immediately find the way the next time we trace the same path? This year’s Nobel Laureates (John O’Keefe, and May-Britt Moser and Edvard I. Moser) have discovered a positioning system, an ‘inner GPS’ in the brain that makes it possible to orient ourselves in space, demonstrating a cellular basis for higher cognitive function.”3
Although scientists are still making beautiful images today, the metaphors used to describe the brain have shifted—from a biological description of the complexity of the brain to a technical and dynamic description of its functioning.
Navigation: Biological and Technical
In 1994, I produced what seems to have been the first artwork made with GPS as a drawing tool. Very few people had heard of GPS in 1994, although it was being heavily marketed at the time by a few companies seeking to spin off this military-first technology as “the next utility.” The exhibition, “You Are Here: Information Drift,” at Storefront for Art and Architecture, was about the architecture of the system that makes digital maps. It highlighted the fact that you have to receive a signal from a satellite in outer space in order to locate yourself, by way of a GPS receiver, on Earth. On a screen in the exhibition, the feed from a receiver on the roof showed a point that, with each refresh, moved around, even though the receiver hadn’t. I wrote about, and displayed the stupidity of, a machine which, by way of communication with a stationary receiver, simply kept drawing the same point every second, never getting tired of jumping around within a range of uncertainty.
The following year I was commissioned to implement You Are Here again for the opening of the museum of contemporary art in the old barrio of Barcelona (MACBA), which was designed by Richard Meier as an all-white and glass building according to a highly rationalist grid. I carried a GPS receiver while walking on the roof to trace the letters of the word “MUSEU.” Meier’s modernist grid provided a frame in which to locate these digital letters. As I walked across the gridded roof, I wondered about the so-called perfection of Meier’s grid compared with this dynamic mapping tool which could not get the lines quite as straight as he would have liked. The idea behind this was that maps mostly point to the coordinate system of the map itself, and not of the city or its building.4
Twenty years later, in 2014, I was working on a project about mapping the human brain and modeling its “connectome.” Our team collaborated with neuroscientists discussing how to represent this networked model, alongside many other models of the brain, to a broad public audience that might not understand the complex ways in which scientists write about their own work. It was a happy coincidence, then, when the three scientists, O’Keefe and the Mosers, were able to explain their discoveries so clearly through the metaphor of the “inner GPS.” By 2014, no one needed to ask what “GPS” stood for.5 For me, however, this metaphor was important to unpack.
The three Nobel scientists had moved from a discovery of what O’Keefe had called “place cells” in 1970 to what the Mosers started calling “grid cells” in 2005. Both scientific and popular reporting about their Nobel Prize for the discovery of these cells referred to the dynamic cognitive maps that both people and animals produce and reproduce whenever they navigate. Accounts for the general public were mostly fascinated by the idea that the brain had a little (GPS) machine in it. Scientists, however, were talking about navigation and spatial memory, and how humans and animals are able to remember space at all. It's not obvious how this happens, but they argued that it’s a necessity of existence. According to the Mosers, “The ability to figure out where we are and where we need to be is key to survival. Without it, we, like all animals, would be unable to find food or reproduce. Individuals—and, in fact, the entire species—would perish."6
It is important to keep in mind that GPS is a set of machines working together in real time, which means that it does not remember anything. It is simply a system that connects satellites to receivers and identifies momentary locations in longitude and latitude. It is only when GPS is connected to other devices, like phones, cars, or control centers, that the information it generates can be stored and transformed into data as a spatial memory, with a date and time stamp.
Place Cells and Grid Cells
Despite the discovery of place cells and grid cells, it is not that easy to understand how our brains visualize the spaces we traverse and actually navigate through them. In biological terms, the cells in our brains—neurons—communicate in very particular ways. It is an axiom of neuroscience that “neurons that fire together, wire together,” which is to say that, by sending electrical signals across voids called synapses, they form specific connections that create networks in the brain.
We owe the notion of “cognitive maps” to the behavioral psychologist Edward Tolman, who coined it in 1948 to explain the conceptual and spatial diagram created in the brain of a rat to remember its way through a maze. Tolman recounted the analogies used by his students in debating how to explain the rats’ behavior moving through a variety of mazes. If the brain was like a “central office,” what happened in that office? Was it a “stimulus-response” situation, like “a complicated telephone switchboard” where “incoming calls from sense organs” were answered with “outgoing messages to muscles”? No: Tolman asserted that “the central office itself is far more like a map control room,” where “incoming impulses are ... worked over and elaborated ... into a tentative, cognitive-like map of the environment.” Not call and response, but “routes and paths and environmental relationships.” 7
Two decades later, John O’Keefe found the actual cells in the brain that helped form these cognitive maps and called them “place cells.” He located them in the hippocampus, the physical part of the brain associated with memory. Navigation through cognitive maps in these experiments was related to learning and remembering space. Place cells fire when you are at a particular location and encode information sequentially, meaning that one can retrace a particular journey by going through a list of “place memories” in sequence. For instance, in New York City, a rat leaves the basement, turns left, and scuttles to the fire hydrant, knowing that facing this particular way means finding a pile of garbage bags—and hence, that it is headed in the right direction for dinner. As the Nobel Committee described it, “The place cells report the position of the rat and build up an inner map that represents a mental picture of the environment. O’Keefe suggested that memory of a place may be stored as a specific combination of place cell activities.”8
In 2005, Edvard and May-Britt Moser looked outside of the hippocampus to understand how this worked. This led them to the entorhinal cortex of the brain, what Edvard called “unknown territory.” Edvard says: “We put electrodes in that part of the brain in rats and recorded activity. In an environment where the cells are very active you will hear this really high-[pitched] popcorn sound and this allowed us to record patterns.”9 The patterns they recorded formed hexagonal grids. They described this grid as “an empty map, like longitude and latitude without the map, just a coordinate system.”10 This strange but familiar formulation implied that the grid had no knowledge of an actual place, or what it looked like—only that the animal had a position and a speed. They said, “maybe you can say that the [grid cells’] similarity is to a GPS is that the system uses the rat’s position and movement [and] speed, but doesn't care about what its environment looks like. It only uses the change in position, just like a GPS.”
As You Are Here made clear, GPS will not stop drawing points until you turn it off. According to the Mosers, the same goes for our brains: they just keep drawing and redrawing maps like coordinate systems, relative only to a starting point—a kind of dead reckoning, where only speed and direction, not landmarks, are relevant. The Mosers did not forget about O’Keefe’s place cells, however. Reiterating the importance of the networked model of the brain, they explain that there is communication between the entorhinal cortex and the hippocampus once you have a specific memory. This is the memory of landmarks, of the environment, of everything that is layered on top of the abstract coordinate system. The Mosers explained, “of course the difference in the navigation system in the human brain (as compared with GPS) is that it's so much more. This area that contains the grid cells happens to be the same area that is first damaged in many Alzheimer’s patients.”11 Bringing this theory to a form of practice, other scientists studied the brains of London black-cab drivers in 2006, and discovered that the legendary three year-long “knowledge” test, combined with their years of experience driving taxis, enlarged their hippocampus. When they retired, their hippocampus shrank.12
The Mosers’ work identified and expanded the vocabulary of types of neurons in the brain—grid cells, place cells, border cells, and edge cells—which are all activated in different parts of the brain and communicate with one another to create a neural spatial network that forms dynamic cognitive maps to varying degrees of abstraction and specificity. Only place cells remember and learn with any specificity about an environment, while grids cells and the rest are empty of specific environmental content. Like GPS receivers, they just register movement independent of any specific environment. The Nobel press release continues, “[Grid cells and place cells] have opened up new avenues for understanding other cognitive functions such as memory, thinking and planning.”13
Revisiting Cognitive Mapping
This new science of the mind is very popular among architects and planners today, particularly with a set of designers and critics who claim to discover in neuroscience a set of fundamental human traits, and who use them as a standard to criticize the current state of the built environment. For example, look at the pop science which camouflages a distinctive ideological position in a 2015 Guardian article about so-called “conscious cities,” where the authors label an image of Times Square as “cognitive overload,” or an image of a cul-de-sac as “suburban blues” or a “numb landscape.”14 These cliches are bad instrumentalizations of the many more compelling things architects can do with neuroscience, ones that don’t hide from the radicality of scientific discoveries. Instead, we might ask how questions of orientation, location, and navigation, are relevant to the fundamental questions of learning and curiosity, and therefore memory—familiar themes in the canons of architecture and urbanism.
In 1960, Kevin A. Lynch picked up on Tolman’s behavioralist concept of cognitive maps in his seminal, if problematic, book, The Image of the City. He interviewed people (thirty people in Boston and fifteen people each in Los Angeles and Jersey City—three cities in total) and asked them to draw “mental maps” of their city as they remembered it. In the book, he focused on the “legibility” of the American city and proposed that “in the process of way-finding, the strategic link is the environmental image, the generalized mental picture of the exterior physical world, that is held by an individual. This image is the product both of the immediate sensation and of the memory of past experience, and it is used to interpret information and guide action.”15
Lynch developed his theory of how urban designers could have an influence on memory in our brains by proposing that these mental maps of the city could be reinforced by specific design elements—paths, edges, districts, nodes, and landmarks—in order to be more effective in terms of navigation and sense-making—and to contribute to what he called “good city form.” In order to do this, Lynch actually talked about the illegibility of the city, particularly Los Angeles and Jersey City. When you need wayfinding signage, mental maps are not formed, and, he thought, urbanism fails. Architects and urban designers have used his principles worldwide since their first publication. Technologists have been equally influenced by Lynch’s navigation guidelines, but symptomatically as metaphors of physical space inside of the digital. Human-computer interface designers, learning from Don Norman and his “Design Thinking” methodology, use Lynch’s universal concepts as the most common and necessary metaphors of space in navigating complex digital interfaces, without questioning the principles or his method of defining them.16
In Fredric Jameson’s reconsideration of Lynch in his equally influential “Postmodernism” essay, he rejected the premise that these “cognitive maps,” formed with such simple legible images and so few criteria, were either necessary or universally shared. Writing in 1984, he argued that “Lynch suggests that urban alienation is directly proportional to the mental unmappability of local cityscapes. A city like Boston, then, with its monumental perspectives, its markers and monuments, its combination of grand but simple spatial forms, including dramatic boundaries such as the Charles River, not only allows people to have, in their imaginations, a generally successful and continuous location to the rest of the city, but in addition gives them something of the freedom and aesthetic gratification of traditional city form.” Jameson proposed that contemporary urban space was too complex simply to be “mapped” in this way, and instead of taking legibility as a given need, claimed that it generated “something like an imperative to grow new organs, to expand our sensorium and our body to some new yet unimaginable, perhaps ultimately impossible, dimension.”17 Think about those London cab drivers.
Jameson probably did not read anything about grid cells, but might have predicted their existence in a way that is compatible with complex neuroscience as a multi-dimensional view of cities. Multi-dimensionality is at the core of neuroscience today; to enable people to see neural networks implies limiting their dimensions, which means hiding some of them, so that ideas about complex networks might be explained. But for Jameson, as an analyst of late modern capitalism, coordinate systems can never be empty in relation to the places we live in: they are the outcome of social and economic forces. Maps have agency in the formation of space. In the last paragraph of his essay, Jameson’s challenge is that “the political form of postmodernism, if there ever is any, will have as its vocation the invention and projection of a global cognitive mapping, on a social as well as a spatial scale.”18
Jameson could not have been more accurate, as money, people, and things have moved around the world with ever-increasing speed in the time since his essay first appeared. Maps that visualize global data have made visible the scale of so many patterns. Geographers, cartographers, and data visualization artists have knowingly or unknowingly taken up his challenge and created global cognitive maps, while also critiquing the ways in which data is deployed to describe the ways in which people, animals, and environments are situated. In a world where the digital has in many ways colonized the physical, cognitive mapping deserves to be retheorized. But is cognitive mapping still a useful category for describing the ways in which we understand and picture the world outside of ourselves?
Navigate With Caution: Neural Networks and Artificial Intelligence
In “Parascientific Mediations,” Ranjodh Singh Dhaliwal helps us locate (as it were) the question of navigation within the broader topic of neural networks, both biological and artificial. He writes: “On the one hand is the neuron … colloquially understood as the thinking cell in most animal species. On the other hand is the artificial neuron, understood analogically as a biomimicry instantiated in silicon computational hardware. Both figures, in this story, are separated by almost a century; this gap is not a historical given—the ties between a neuron and an artificial neuron get inaugurated right in the middle of the twentieth century, after all—but instead a choice made here to illuminate some specific transhistorical features of, and maneuvers in, thinking about thinking.”19
Starting his 100-year history with Cajal and Golgi, Daliwal ends with John Hopfield and Geoffrey Hinton, who received the Nobel Prize in Physics for their work on machine learning, or artificial intelligence, in 2024. Hopfield and Hinton’s work is explained by the Nobel Prize press release, as it is by Dhaliwal, through the brain/computer analogy: “When we talk about artificial intelligence, we often mean machine learning using artificial neural networks. This technology was originally inspired by the structure of the brain. In an artificial neural network, the brain’s neurons are represented by nodes that have different values. These nodes influence each other through connections that can be likened to synapses and which can be made stronger or weaker. The network is trained, for example by developing stronger connections between nodes with simultaneously high values.”20
The key word here, which should not be taken for granted or overlooked, is “inspired.” The technology of machine learning using artificial neural networks is conceptualized by the scientists as an abstract, machinic model made up of artificial neurons. While the biological neurons known as grid cells—the empty hexagonal grids—were described metaphorically by the Mosers as “the brain’s GPS,” Hopfield and Hinton’s models of machine learning invert the analogy, treating artificial intelligence as a technological analogue of the biological brain.21 Artificial neural network charts label the inputs and outputs of these “brain-like” machines. Their inner workings, however, are labeled as “hidden layers,” which represent the weighted and biased priorities driving their calculations.22 Colloquially known as “black boxes,” these hidden layers are used in training what is generally described as the “next machine”: an updated AI model even more intelligent than the last one, with new inputs, new hidden layers, and new outputs.
These brain-like artificial networks learn things through the dynamic networks which build them, phenomena embodying multiple intelligences translated into data: biological, ecological, technical, imaginative, predictive, creative, and destructive. In the most common models, such as the large language models underlying generative AI, it is often our own data which is providing the intelligence for a collective (or corporate) artificial brain. The actions we take on our smart phones—the emojis we express, the images we post, the searches we undertake, the articles we read, the things we click on and buy—all feed the artificial neurons in the network, even their hallucinations.23 These activities help the artificial brain to predict our next move in the network, in order to guide its own action.
The boundaries between biological and technological brains are blurring. If our biological brains have an inner GPS and our technical infrastructures are embedded with artificial neurons, cognitive maps are constantly being generated. But what kinds of mental maps are they drawing? This brings us back to the idea of dead reckoning, the sixteenth-century nautical term for navigating without landmarks (on Earth or in the heavens) from a known starting point to a particular destination. With no landmarks, you can only ascertain your direction (compass), and track your speed over time (distance). But with this information, you can “reckon” (calculate) your position without external references. However, metaphorically speaking, dead reckoning can mean many things: drifting off course, using one’s internal compass, moving from a known past to an uncertain future, or feeling your way forward when trusted systems of knowledge collapse.
The cognitive mapping activities of neural networks, be they biological or artificial, picture one’s uncertain location, or rather, too many uncertain destinations, in dynamic networks. In the biological brain, grid cells are structured like a geographic coordinate system (longitude and latitude), vacant of any content; they only “reckon” or calculate by knowing our speed, borders, edges, and direction. Place cells, however, record specific things and places—memory—enabling us to learn, desire, and be curious, as well as avoid and fear. While some artificial neural network models do form memory, find patterns, and make predictions, moving through any kind of neural network—either biological or artificial—means navigating high-dimensional space.
Scientists often diagram high dimensionality by “projection” into lower-dimensional spaces, like a shadow play where three-dimensional forms are flattened onto two dimensions for graphing. But what do these analogical brains learn in the process of being trained, being strengthened and weakened? Their memory and predictions have agency; their hidden layers are designed and controlled by the ideologies that have trained them. As always, the tools we use enable and limit what we can do. When the boundaries between the tools and their ostensible users become blurred, we need to learn to navigate cautiously, for a lot is at stake between the biological, the technical, the artificial, and the political.
Carl Schoonover, Portraits of the Mind, Visualizing the Brain from antiquity to the 21st Century (New York: Abrams, 2010), 7.
“Speed Read: Exposing the Forest,” The Nobel Prize, September 16, 2009, ➝. Italics mine.
Nobel Prize in Physiology or Medicine 2014 press release, The Nobel Prize, October 6, 2014, ➝.
This work is extensively documented in my book: Close Up at a Distance: Mapping Technology Politics (New York: Zone Books, 2013) 58–85.
Lawrence K. Altman “Nobel Prize in Medicine Is Awarded to Three Who Discovered Brain’s ‘Inner GPS’,” October 06, 2014, ➝.
May-Britt Moser and Edvard I Moser, “WHERE AM I? WHERE AM I GOING,” Scientific American 314, no. 1 (January 2016): 26–33.
Edward C. Tolman, “Cognitive Maps in Rats and Men,” The Psychological Review 55, no. 4 (1948): 189–208.
Ole Kiehn, Award Ceremony Speech for the Nobel Prize in Physiology or Medicine 2014, The Nobel Prize, December 10, 2014, ➝.
Edvard I. Moser and May-Britt Moser, “A Journey Into Entorhinal Cortex,” interview, posted January 25, 2011, by NTNU University, YouTube, ➝.
Moser and Moser, “A Journey Into Entorhinal Cortex.”
Moser and Moser, “A Journey Into Entorhinal Cortex.”
Mo Costandi, “The Brain Takes a Guided Tour of London,” Scientific American, March 21, 2017.
Nobel Prize in Physiology or Medicine 2014 press release.
Itai Palti and Moshe Bar, “A Manifesto for Conscious Cities: Should Streets Be Sensitive to Our Mental Needs?” The Guardian, August 28, 2015, ➝.
Kevin Lynch, The Image of the City (Cambridge, MA: MIT Press, 1960), 4.
Don Norman, The Design of Everyday Things (Basic Books, 1988).
Fredric Jameson, “Postmodernism, or, the Cultural Logic of Late Capitalism,” New Left Review, no. 146 (July/August 1984).
Jameson, “Postmodernism,” 92.
Ranjodh Singh Dhaliwal et al., Neural Networks (Minneapolis: University of Minnesota Press, 2025), 55.
Nobel Prize in Physiology or Medicine 2014 press release.
Moser and Moser, “WHERE AM I?,” 32.
For a good explanation of neural networks, see Fangfang Lee, “What is a Neural Network?” IBM, September 10, 2025, ➝.
Beth Coleman, Reality Was Whatever Happened: Octavia Butler AI and Other Possible Worlds (Berlin: K. Verlag, 2023).