Neurons were first extensively illustrated by Cajal in ~1890. When he looked at a tissue of the brain, here’s what he painted:
At that point in time, we didn’t know that galaxies other than the Milky Way existed. We didn’t even know that atoms consisted of electrons and protons.
Fast forward to today. We now know that our galaxy is simply one of the 100 billion galaxies out there. We have not only discovered protons but now know that they aren’t even fundamental particles, quarks are. We’ve imaged black holes, detected gravitational waves. We have an accurate account of the universe’s origin starting from the very first moments. In short, we’ve pretty much nailed our understanding of the physical universe since the time Cajal first looked at the brain.
Cajal would have expected that the field of neuroscience would have made an equivalent amount of progress as physics. But, no, reality turned out to be different. We are no better than him in 1890 in answering how the cells in our brain give rise to our rich subjective experiences and emotions. We don’t know what causes the redness of the red and the fruity smell of vanilla.
Why we’re nowhere close to understanding the brain?
It’s definitely not due to a lack of effort or imagination. Over the years, many scientists have put forward their proposals about how the brain works. The field of neuroscience continues to expand and more papers are published each year as compared to previous years.
But, despite all this effort, our understanding of the brain is stuck because we lack high-quality human brain activity data along with descriptions of subjective experiences corresponding to that activity.
Let me expand.
Current technologies for collecting brain activity data are all deficient in various ways. An idealized technology for collecting brain activity will be able to read and write neural spike activity individually for all 86 billion neurons spanning the 3D volume of the human brain. Right now, due to ethical reasons, non-invasive methods are used for experimenting on humans and they have limited spatial or temporal resolution.
For example, fMRI can localize brain activity to a tiny region but since it uses blood flow as a proxy for neural activity, it isn’t a realtime indicator. Activity detected by fMRI changes on the order of a few seconds while neurons fire on the order of milliseconds. Similarly, EEG can detect neural activity changes on the order of milliseconds but since it detects aggregated electric fields across clusters of neurons, it’s impossible to localize the activity to a region in the brain (let alone a single neuron).
Invasive methods such as microelectrode arrays provide individual neural spike data with a high temporal resolution, but you can only map a tiny region of the brain with it (e.g. the FDA approved Utah array can read activity from 100 neurons while the human brain has 86 billion). Moreover, you can’t use invasive methods on humans and animals can’t speak about their conscious experiences. So, triangulation of neural activity to a specific conscious percept (e.g. the smell of rose or the color red) is done indirectly at best.
Neuroscientists simply don’t have access to high-quality neural activity data for human brains. Contrast this with physics where they have the Hubble telescope, LHC, LIGO and other massive experiments generating a constant stream of high quality data.
Without high-quality and reliable data, no science can progress. All we can have without data are hypotheses (which we have plenty of) but no certainty.
What’s needed to accelerate the science of consciousness?
To help build a science of consciousness in this century, we need to invest in developing non-invasive individual neural activity recording and stimulation technologies that work in live human brains.
Breaking this down further:
- Non-invasive because there’s no way anyone will be okay with humans risking brain surgery for invasive methods just for open-ended research. It’s not ethical nor legal to open a healthy human’s brain in order to collect neural activity data. Currently, our only chance to do such kind of studies are non-pharmacologically treatable epilepsy patients (or use animals). Epilelpsy patients are limited in numbers and the brief window neurosurgeons give to researchers for studying neural activity isn’t nearly enough in quantity or quality to develop a full science of consciousness.
- Individual neural activity because, frankly, we don’t know any other fundamental unit of brain activity apart from neurons. Even if it turns out that, say, groups of neurons play a key role in consciousness and not neurons individually, we still need individual neuron data to be able to reach that conclusion. My only worry with a focus on individual neural activity is that if the important entities for consciousness turn out to be sub-neuronal (like dendrites or synapses), we’d get stuck again. But before we worry about that, let’s focus on neuronal activity, which is our current best bet for being the fundamental unit of the brain.
- Recording and stimulation because we need to test our developed theories by directly and precisely manipulating brain states and checking whether the caused subjective experiences match what different theories predict. Such interventionist approach requires direct stimulation of neural activity to put brain networks in a desired state of firing. Right now, we rely on indirect evidences where patients get tiny parts of their brain removed for epilepsy treatment and we try to deduce in what ways the patient behaves and experiences differently post-treatment. But what we learn from such instances is crude (eg this location is for reading faces, or stories memories) and the exact mechanisms of the brain elude us.
- Live human brains because animal experiments are unethical and they can’t efficiently or unambiguously communicate their subjective experiences to us. Training animals to do a task relevant for the experiment can easily take weeks and months and, even then, the interpretation of what they’re experiencing is indirect at best. We need human subjects to be able to communicate to us in detail what they’re experiencing so we can map their neural activity to those experiences which is what’s required for developing a science of consciousness.
Developing technology with the attributes listed above is going to be extremely challenging, but no known physical law prohibits it.
Ultimately, if we want to understand how the brain generates consciousness and how our world comes into existence, we better get down to advancing technology for neuroscience.
This essay started out its life as a raw note in my knowledge garden.