The Neuroscience Revolution: A Long Road Ahead

[Read Part One] [Read Part Two] [Read Part Three]

Am I too optimistic? Despite remarkable technological advancements and an explosion of neuroscientific data, we remain profoundly distant from truly understanding how brains function. In October 2023, neuroscientist Per E. Roland of the University of Copenhagen published an article in Frontiers in Systems Neuroscience that presents a sobering assessment of the current state of neuroscience, arguing that the primary obstacles to progress are not primarily technical but conceptual in nature. As Roland directly states: "The reason why neuroscientists do not understand how brains and central nervous systems work is that there is no theory of brains and central nervous systems."

According to Roland, the fundamental issue is that neuroscience lacks a comprehensive theory of the brain. While we have made significant progress in understanding cellular biology and molecular mechanisms, we still cannot explain how neurons collectively operate to produce the vast repertoire of brain functions. Roland defines what such a theory would require: "A scientific theory of a central nervous system (CNS) is an experimentally based general set of explanations of how the elements in a CNS interact at all scales of observation, i.e., from the molecular to the macroscopic scale." He notes that while molecular neuroscience is guided by the theory of molecular biology, "at the cellular, and especially supracellular scales of observation, neuroscience is far from having a guiding theory." This comprehensive framework simply does not exist.

The article identifies several major conceptual barriers that hinder progress. Perhaps most fundamentally, neuroscience lacks concepts rooted in experimental results that can adequately explain how neurons interact across different scales. Instead, the field relies heavily on borrowed concepts and analogies from other disciplines like engineering, physics, and psychology. Roland states bluntly: "These borrowed concepts are used as analogies in neuroscience. But the borrowed concepts are not tailored to explain (more complex) biological systems such as brains. Logically, analogies cannot and do not explain how neurons collaborate to achieve the whole repertoire of CNS activities."

He emphasizes that "if we remove all analogies and metaphors as attempts to explain brain mechanisms in neuroscience, will we lose understanding of brains? Logically, the answer is no." Without proper neuroscientific concepts, the field struggles to form appropriate hypotheses, design effective experiments, analyze data meaningfully, and interpret results accurately. Roland identifies this as a major frontier: "Develop concepts strongly rooted in experimental results explaining how neurons (and glia) interact at all scales."

Another significant conceptual obstacle is the prevalent assumption that time operates as an independent variable for brain processes. Roland presents compelling experimental evidence that challenges this view, showing that neural activities often evolve with time and space as mutually dependent variables. He notes: "Claiming that all activities in brains all evolve according to external clock time only (i.e., with time as the independent variable) is a strong hypothesis that can be proven wrong." For example, studies reveal that small groups of neurons fire in specific spatial sequences that cannot be explained as temporal patterns: "Subsets of 2–6 neurons elicited from 2 to 6 spikes always in the same spatial order... These results cannot be explained as a brain activity using clock time as the independent variable."

Similarly, the brain's processing of visual information involves predictive spatial dynamics rather than purely temporal processing. Roland concludes: "These examples demonstrate that all brain activities cannot be explained as evolving with clock time as the independent variable." He suggests spatial dynamics—the propagation of activity variables through neural networks—may provide a more appropriate framework for understanding brain function, defining it as "changes in activity variables... propagate through the network of neurons that makes up a central nervous system."

The article also challenges conventional distinctions between task-related and spontaneous brain activity. During experimental tasks, Roland notes that "40% of the neurons in the brain and mesencephalon may increase their spiking, and up to 20% of neurons decrease their spiking, whereas the remaining 40% of the neurons do not change their ongoing spiking." Additionally, "a large proportion of neurons (up to 40% of all neurons) may not spike at all." This highlights how difficult it is to separate task-related from intrinsic activity, leading Roland to identify another conceptual frontier: "Separate self-organized intrinsic activity in CNS from task dependent activity."

Determining whether brains are primarily driven by external stimuli or are largely autonomous systems is another unanswered open question. Evidence suggests that most neural activity may be self-generated: "Even in primary visual and auditory cortical areas, only 5%−15% of the spikes carry information about the surround... These results are well known and indicate that 85%−95% of the spikes in a brain are autonomous." Roland further notes: "Until recently, neuroscience has been mainly oriented to explain how changes in the surrounds and behavioral conditions change transmembrane currents... Recently, there is accumulating evidence contesting this view that spiking and post-synaptic dynamics in brains are predominantly externally driven." This suggests that neuroscience may need to fundamentally reconsider how it conceptualizes brain function.

Beyond conceptual obstacles, Roland identifies significant experimental challenges. One fundamental issue is the lack of appropriate baseline or reference conditions for measurements. Trained animals and humans are never truly naïve during experiments—they exhibit preparatory spatial dynamics before trials even begin, making it difficult to establish a true baseline for comparison. Roland states: "Awake-trained animals and humans are not naïve. In contrast, they are specifically engaged in performing the task prior to the experimental trial. Prior to the experimental trial, spatial dynamics evolves in the brain stem, hippocampus, basal ganglia, and cortex." This leads him to ask: "Which reference should measurements from brains have?" The traditional solution of using a "rest condition" is problematic because "the assumption that this 'rest state' is stationary and valuable as a reference for trials done immediately before or after the rest measurement is most likely false."

Additionally, the practice of averaging across neurons, trials, or brain areas obscures the underlying spatial dynamics. Roland asserts: "Averaging across neurons, single trials, single areas, or other CNS structures hides the underlying spatial dynamics," suggesting that single-trial designs and analyses are necessary for capturing the true nature of brain activity. He argues: "The assumptions underlying temporal and spatial averaging, multi-trial statistics, and statistical independence of trials are most likely wrong. So, neuroscientists are forced to design single-trial experiments and analyze single trials statistically."

Technical limitations also persist. While modern techniques allow for simultaneous recordings from thousands of neurons, they remain limited in revealing the spatial dynamics of processing in axonal branches and dendrites, connecting this activity to appropriate neurons, and extending comprehensive recordings to primates. Roland notes that "spike recordings do not reveal the type of neurons involved (excitatory glutamatergic, inhibitory GABAergic, and glycine-ergic sub-types). Moreover, extracellular spike recordings are blind to the dendritic contributions." The interdigitation of dendrites from thousands of neurons creates immense challenges: "The local interdigitation of dendrites from thousands of neurons... implies that post-synaptic transformation by individual neurons cannot be resolved with one-photon, two-photon, or three-photon optical recordings, because it is difficult to match the active dendritic branches with the right neuron." This leads to a technical frontier: "Reveal the spatial dynamics in axonal branches and of synaptic and dendritic processing and connect this to the appropriate neurons." Additionally, Roland notes that "including primates is so far out of reach for comprehensive spatial dynamic recordings."

The interpretation of experimental results presents further challenges. The lack of reference conditions, continuously changing neural activity, absence of clear initial states, and spatial-temporal nature of brain dynamics make traditional cause-effect analyses inappropriate. Roland summarizes these interpretational obstacles like this:

"The continuously changing spiking and changing transmembrane currents everywhere in a CNS implies that one cannot apply a classical cause-effect analysis: if A, t1 then B, t2."

"Central nervous systems, in contrast to complex dynamical systems, have no clear initial state definition, neither locally nor globally. This implies that we cannot explain the future states of the system from local or global initial states."

"Time is probably not an independent variable for CNS operations. In a CNS, dynamics are space and time dependent, i.e., spatial dynamics. This implies that pooling data from different neurons or locations and temporal and spatial averaging destroy the spatial dynamics."

The limitations of dynamical systems theory, the difficulty of separating task-related from intrinsic activities, and the current restriction to observing spatial dynamics in only discrete parts of the CNS all complicate the reliable interpretation of results. Roland poses three theoretical frontiers: "How can we reliably interpret our results?", "How can we reliably explain our results?", and "How can we start to make theories of brains?"

Roland proposes several frontiers for advancing neuroscience as pathways toward a better understanding of brain function. He emphasizes the need to "develop concepts strongly rooted in experimental results explaining how neurons (and glia) interact at all scales," rather than relying on borrowed analogies. The field must also work to "build a brain model with modifiable, but everlasting ongoing changes of membrane potentials and spiking like that in mammalian brains," since current models lack the continuous ongoing activity characteristic of real brains.

Roland stresses the importance of understanding "which (biophysical) mechanisms determine how far depolarizations and spiking spread in CNS," moving beyond oversimplified circuit models. At the cellular level, neuroscience needs to "understand the local and global in vivo processing in dendrites of single neurons and their consequences for emission or withholding action potentials," recognizing that dendritic processing is largely neglected in current models. The field must develop methods to "separate self-organized intrinsic activity in CNS from task dependent activity" and "measure regulation of CNS autonomy," acknowledging the brain's largely self-generated activity.

He also emphasizes the importance of discovering "principles for how intrinsic activity in brains emerge" and the need to "reveal the spatial dynamics of subcortical structures at all spatial scales," as current understanding focuses heavily on cortical processes. Finally, he advocates using the framework of spatial dynamics to "find principles of interactions of neurons at all scales of observation," potentially providing a conceptually sound basis for future theories.

In his conclusion, Roland defines what a scientific brain theory would require: "A scientific brain theory would be an experimentally based general explanation on how the elements in brains interact at all scales of observation under all conditions. A theory must serve as a conceptual structure in which gaps of knowledge and inconsistencies can be isolated. It must offer rules and coherent explanations, to some extent encompassing different scales of observation."

Roland's analysis presents neuroscience as a field still in its infancy in terms of truly understanding brain function: "Despite a century of anatomical, physiological, and molecular biological efforts scientists do not know how neurons by their collective interactions produce percepts, thoughts, memories, and behavior. Scientists do not know and have no theories explaining how brains and central nervous systems work." He concludes that "the main obstacles for systems neuroscience seem to be conceptual. That is lack of concepts rooted in solid experimental results, unnecessary assumptions, and focus on analogies from other disciplines." Progress will require not just more data or better technology, but fundamentally new concepts and approaches that can accurately capture the unique nature of neural systems.

Given the challenges highlighted here, it's a thoughtful question to ask to what degree neuroscience shows promise for understanding cognition, behavior, and well-being at this point in time. Roland's critique offers a serious counterpoint to my enthusiastic perspective on neuroscience's current state and future potential as expressed in my previous entires where I describe neuroscience as "the most powerful force in the world no one is talking about" with "more potential to revolutionize human behavior and cognitive well-being than anything else available to us."

Roland would certainly challenge my assertion that "for the first time in human history, we are discovering how the brain operates." His analysis suggests we're much further from such understanding than my blog entries imply. He would likely view my characterization of neuroscience as witnessing "the ignition phase of a knowledge explosion" as prematurely optimistic given the fundamental conceptual obstacles he identifies.

Yet I maintain that these seemingly contradictory perspectives are not entirely at odds. Roland focuses primarily on theoretical understanding, our ability to explain how neurons collectively produce the full spectrum of brain functions. My focus is more on the practical advances, specific insights, and actionable knowledge that neuroscience continues to generate. Both perspectives have validity.

Roland's critique doesn't invalidate the real and quite recent progress neuroscience has made in understanding specific brain mechanisms and their relationship to behavior and well-being. The research I highlight in my blog entries, from Huberman's work on dopamine and motivation to Keltner's studies on awe, represents valuable advances in our understanding of brain function, even if they don't yet constitute a comprehensive theory of how brains work. These discoveries have practical implications for enhancing human well-being today, regardless of whether we fully understand how they fit into a unified framework of brain function.

Moreover, the multidimensional approach of modern neuroscience that I celebrate in my previous three posts is likely a necessary foundation for any future integrated theory. Without understanding brain function across diverse levels and contexts, we couldn't possibly develop the comprehensive theory Roland finds lacking. The diverse research approaches aren't a problem but rather a strength and prerequisite for eventual theoretical integration. Roland might see everything I've cited as fragmentary, but the potential is real and the results in everything I've discussed are extremely positive. You have to build upon something solid, even if you don't know how everything ties together.

Perhaps the most balanced view is that neuroscience has indeed "launched the rocket" toward better understanding the brain and human well-being, but leaving the launch pad does not mean we are out of Earth's atmosphere yet and we are nowhere near our destination. We have powerful methodologies and technologies, have identified numerous brain mechanisms that correlate with specific aspects of cognition and behavior, have developed effective interventions based on neuroscientific insights, and have begun to understand how various factors directly affect brain function and well-being. A lot of this was not even known at the time Roland authored his insights.

However, as Roland's critique makes clear, this rocket still faces significant challenges. The diverse findings I've cited are not integrated and certain assumptions about how to study the brain may be fundamentally flawed. There are unresolved questions about brain autonomy versus external influence, while appropriate experimental protocols and analytical methods are still being debated.

To foster legitimacy it is important to temper my optimism, to be enthusiastic about concrete advances while clear-eyed about conceptual limitations. Realistically, neuroscience has made remarkable progress in understanding specific brain mechanisms and their relationship to behavior and well-being, even as it struggles with fundamental conceptual challenges that limit our theoretical understanding. My optimism about neuroscience's potential to enhance human well-being remains well-founded, but Roland aptly reminds us that fully understanding how brains work is longer and more demanding than I might hope.

But we do understand our brains more today than we did a decade ago. I find it unlikely that this pace of discovery will slow in the decade to come, even if the metaphysical understanding Roland says we don't have remains hidden for the moment. He doesn't suggest that a comprehensive understanding of the brain is impossible, just that we're facing significant conceptual obstacles. He identifies specific frontiers where progress is needed and proposes spatial dynamics as a promising framework, showing he believes advancement is possible. This critique is actually healthy and will lead to more refined methodologies.

The explosion of research across multiple domains of neuroscience, from genetic mechanisms to neural circuits to behavior, has yielded concrete insights that weren't available before. We have a better understanding of how specific brain circuits regulate fear responses and risk assessment, the neural basis of empathy and social cognition, the impact of experiences like awe on brain function, the neurobiological pathways of adaptive learning, among a multitude of other solid understandings. We've answered a lot of hows even if we don't know all the whys.

Onward!

(Assisted by Claude.)