The vivid redness of a rose, the soothing sound of the waves and the sweet blissful flavor of chocolate in your mouth—these are all sensations that we are familiar with. From a scientific perspective, these constitute stimuli that the brain can comprehend and manifest into a plethora of emotions that contribute to our subjective experience. Before we can feel the redness of red, however, the brain needs to extract features from the world, put them together and form an internal representation of the outside. This was described by David Marr, during his studies on vision, who believed that the brain should be understood as an information processing system. He described the early stages of visual processing to generate a primal sketch of a scene by extracting features such as lines and curves, much like a rough sketch or basic concept art. This basic sketch is then sequentially built up with more and more complicated elements until a complete three-dimensional model of the surroundings is constructed in the brain. This idea of isomorphism; that a stimuli has a physiological representation in the brain, is taken for granted in modern neuroscience. Many experiments in neurophysiology involve investigating the relationship between experimentally provided stimulus and the internal response the brain generates. This internal response is detected by measuring magnetic or electric activity that brains generate and is funneled into a gamut of complex analyses. Challenging isomorphism is the first step György Buzsáki takes in his book, ”The brain from inside out”, to explain the philosophical and scientific rationale behind his neuroscientific practice.
Buzsáki’s fundamental argument stems from what Stevan Harnard, a cognitive neuroscientist, refers to as the grounding problem. The grounding problem, in the context of neuroscience, states the difficulty of understanding how the meaning of a symbol can be gained from other meaningless symbols. Or as Buzsáki puts it, how can neurons, which are basic elements of the brain, understand the features of the outside world without having any prior information about them? He goes on to describe the brain as a system that is completely blind to the outside world. Inside the brain, there exist only brain signals, therefore, there is no direct means to compare internal signals to what is outside in the environment. The brain cannot see if the changes in brain signals are due to external perturbations or by the brain’s self-organized activity. Without further information the sensory neurons cannot assign any meaning to these stimuli. To elaborate on this point, Buzsáki provides a tantalizing thought experiment. Imagine a camera (a sensory device) on a robotic arm that is connected to a culture of neurons. These neurons are taken from the visual cortex, to make a cybernetic model organism that can be experimentally investigated. Bypassing the technical challenges, consider that the region of these neurons that is stimulated by the camera can be dubbed as the “sensory region”. When a stimulus, such as an image of a rose, is shown to the camera; certain similar patterns may start to emerge in the neuronal activity. There may even be interesting alterations in the synaptic connectivity between neurons. The avid experimenter could then label certain patterns of activity to stimuli shown—possibly hinting towards discovering a particular code of the stimuli’s features. The cultured neurons, however, have no direct access to the outside world and have no way of “verifying or grounding their patterns to the events outside”. For example, for mapping an English-to-Chinese dictionary, you could ground a known English word to the Chinese version (after they have been correlated) to give it the same meaning. However, the current challenge would be more akin to the impossible task of trying to ground English to Chinese words (a language you are unfamiliar with) by using a Chinese-to-Chinese dictionary.
To overcome this problem, the essential mechanism for establishing the ground truth depends on an action-perception arc. Imagine now, that a dedicated region of spontaneously spiking neurons are connected to the robotic arm of the camera that enables them to move it. The neurons in the dish now have two major functions. First, to generate an output to move the camera and second, to respond to the camera’s visual input. Due to the interconnectedness of the neuronal regions, neurons in the “sensory” region receive a bi-variate input: input from the camera’s signals—an agent saying: “this is what I see”—and an input from the motor partners saying “this is what I changed”. In essence, the neurons can now see the outside world as well as the internal computational monologue, compare them both and develop a sense of the environment based on their own initiative. Assigning a goal to this system, such as finding a rose, puts things into a more tangible context. Whenever the camera managed to focus on the rose, via random movements, an evolutionary reinforcing agent (such as dopamine) causes strengthening of the connections between the neurons that caused the motion of the robotic arm to the rose and those that were activated by the visual inputs of the camera. Over multiple repetitions, this modification may increase the probability for the camera to be directed towards the rose. Therefore, a meaningless brain pattern can be linked to an action, which enables that pattern to gain meaning and significance to the organism. Immediately after describing this thought experiment Buzsáki remarks:
The picture I just painted is unfortunately an oversimplification. I did not mean to create the impression that our tissue culture thought experiment can explain the mechanisms of a thinking, feeling brain. It was only meant to illustrate the minimum necessary requirements of a brain-like system
This core idea is explored and elaborated in Buzsáki’s book, with an incredible amount of physiological evidence to support it. More speculative ideas, such as imagination and planning also are important derivatives built upon the central theme of the action-perception arc. The thesis of this book brings into question the entire aspect of the stimulus and representation framework that many neuroscientists, including me, work in. This fundamental thesis is reminiscent of Dale Purves’ main theme in his book, “Brains: how they seem to work”. In his book, Purves discusses how objective reality is not what the brain learns to perceive, rather the only thing that the brain encodes is empirically successful behavior. I distinctly recall Purves remarking on the shortcomings and difficulty of establishing a cohesive theoretical account of Hubel and Wiesel’s landmark experiments mapping receptive fields of the feline visual cortex. As someone who really admires the elegance of the three-stage representation framework of Marr, I still recall the gut wrenching feeling I felt as I went over Purves’ words in his final chapter reading:
When a path in science is pursued for this long without the emergence of a deeper understanding of the issue being addressed, doubts are usually warranted. Since the 1980s, it has become increasingly apparent that neuronal responses to even the simplest visual stimuli are difficult to rationalize in terms of a hierarchy that begins with the detection of image features at the retinal level and ends with feature representation in the cerebral cortex
The inner workings of the brain still remain elusive, and that is precisely why we constantly need to push our thinking beyond the current paradigms. Buzsáki and Purves both have this in common, that they are interested in the rules of the syntax, rather than specific vocabulary. Understanding the operational principles of the brain will require every ounce of creativity and scientific rigor across the disciplines, and as scientists we need to be willing to continuously challenge our current way of thinking. This entails uncertainty, hardship as well as moments of bliss. Our flexibility—or should I say plasticity—will pave the path to inking the intricacies of the brain onto paper.
I would like to thank Katrina Deane for her edits and insights on the article.