(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



Does brain activity cause consciousness? A thought experiment

['Albert Gidon', 'Institute Of Biology', 'Humboldt University Berlin', 'Berlin', 'Jaan Aru', 'Institute Of Computer Science', 'University Of Tartu', 'Tartu', 'Matthew Evan Larkum', 'Neurocure Center For Excellence']

Date: 2022-06

Rapid advances in neuroscience have provided remarkable breakthroughs in understanding the brain on many fronts. Although promising, the role of these advancements in solving the problem of consciousness is still unclear. Based on technologies conceivably within the grasp of modern neuroscience, we discuss a thought experiment in which neural activity, in the form of action potentials, is initially recorded from all the neurons in a participant’s brain during a conscious experience and then played back into the same neurons. We consider whether this artificial replay can reconstitute a conscious experience. The possible outcomes of this experiment unravel hidden costs and pitfalls in understanding consciousness from the neurosciences’ perspective and challenge the conventional wisdom that causally links action potentials and consciousness.

Funding: JA was supported by the Estonian Research Council grant PSG728, European Social Fund via IT Academy programme, and the European Union’s Horizon 2020 Research and Innovation Programme (under the Marie Skłodowska-Curie grant agreement no. 799411). MEL was supported by the European Union’s Horizon 2020 Research and Innovation Programme (785907/HBP SGA1, SGA2, SGA3, and 670118/ERC ActiveCortex) and the German Research Foundation (LA 3442/2-1, EXC 257 NeuroCure, and project number 327654276 – SFB 1315). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Introduction

The idea of stimulating the brain to evoke conscious experiences has a long history in neuroscience [1–4]. Nowadays, brain–machine interfaces [5] encode and decode neuronal activity [6–8] and are routinely used to control neuroprosthetics [9]. Electrical stimulation of sensory brain areas is becoming sufficiently precise to deliver specific content, bypassing sensory organs [10] or diseased brain tissue [11]. Furthermore, it is now possible to evoke a memory by selectively reactivating ensembles of neurons (i.e., the engram) that were naturally active in the animal’s brain in a previous event [12,13] (for review, see [14]). Although brain activity can take many forms, it is almost always associated with the neuronal firing of action potentials. Moreover, the effective use of action potentials in brain–machine interfaces with neuroprosthetics and rehabilitation of neural function [5] suggests that action potentials are the fundamental unit of information in the brain.

In experiments routinely performed in neurobiological laboratories, action potentials are recorded and evoked in single neurons and even in small-scale networks [15,16] using current clamp and voltage clamp techniques. Using these techniques, triggering action potentials at the researcher’s bidding (rather than naturally due to the synaptic inputs) is commonplace and even mundane in a modern electrophysiological laboratory. The rapid development of tools and technologies in neuroscience [17–21] brings the goal of capturing every action potential in every neuron of the brain ever closer [22,23]. To date, the highest number of channels recorded by an electrode array belongs to the Argo system, with 65,536 channels [24]. These technologies provide unprecedented insights into the fine details of brain function. Thus, it is perhaps just a matter of time until newer, more powerful technologies will eventually allow us to solve the mechanics of how the brain works. As we converge on this goal, will we get closer to understanding brain function and, with it, the biological causes of conscious experience?

The fact that there is no commonly accepted definition of consciousness has not prevented researchers from pursuing the neural mechanisms underlying consciousness [25,26]. Here, we took the approach that it is sufficient to identify the target of the investigation rather than strictly define it [27]: Consciousness is the experience of ourselves and the surroundings that fades when we enter deep sleep or under anesthesia (cf [28,29]). In a typical experimental paradigm to study consciousness, a visual stimulus is briefly presented to a participant. The stimulus is constructed to be consciously perceived in some trials and not in others [30–32]. Contrasting neural activity of these 2 types of trials allows researchers to delineate the neural processes underlying consciousness (with some caveats, see [30,31]). The research of the fundamental questions regarding the mechanisms and functions of consciousness has also benefited us with new tools to diagnose disorders related to consciousness [33–35].

Here, we revisited “The Story of a Brain” by Zuboff [36] in light of the advances in neurotechnology and their potential role in unraveling the neural causes of consciousness. We consider the consequences of an experiment where a participant’s brain is manipulated in 3 steps using extrapolated versions of technologies currently within the grasp of neuroscience [37], specifically, voltage clamp and optogenetics. Voltage clamp [38,39] can fully and precisely determine (i.e., clamp) the neuron’s membrane potential. An amplifier computes the current to be injected into a neuron via an electrode such that the neuron voltage matches the experimenter’s “command” voltage. An intelligent offspring of the voltage clamp is the “action potential clamp” [40], which, as its name suggests, clamps the neuron voltage to a previously recorded action potential (Fig 1A and see [41,42]). The replay, which voltage clamps the neurons, is not simply superimposed on the neuron’s activity but rather fully determines it by overriding all naturally occurring voltage changes. Consequently, the role of connectivity, feedback connections [43,44], and information propagation is subsumed by the replay.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 1. Recording and replaying action potentials in the entire brain. (A1) Experimental schematic of the action potential replay in the squid giant synapse as done in [41]. (A2) Action potential recorded from the presynaptic terminal (solid purple line) is set as the “command” voltage. (A3) The action potential recorded in A2 (i.e., the command voltage) is replayed (dotted purple line), and the postsynaptic response was “virtually indistinguishable from that obtained with the original presynaptic spike activation” [42]. (B) A stimulus (green light on a screen) is presented to the participant while all action potentials from each neuron in the brain are recorded and stored for later retrieval. The participant is asked to press a button when she consciously perceives the green light. (C) Step 1: All patterns of the action potential are played back into each neuron using the voltage clamp. The participant presses a button during the replay because the relevant motor neurons are activated. (D) Activity in 2 representative neurons from the participant’s brain; during the recording, neuron a fires at time T 1 and causes action potential in neuron b at time T 2 . Red arrows indicate the direction of cause and effect between the neurons and the controller; Neuron a causes action potential in neuron b, and both are recorded by the controller. E. During replay, neuron a and neuron b fire exactly at times T 1 and T 2 , respectively, as in C, but both action potentials are caused by the replay controller (depicted by the green arrows). In both C and D, the action potentials propagate through the axon (depicted by the red arrow), but in D, they do not affect the firing of neuron b. https://doi.org/10.1371/journal.pbio.3001651.g001

For each step in the thought experiment, we ask whether activating the brain with an artificial replay of previously recorded action potentials would result in conscious perception and explore the possible outcomes. It is impossible to say whether replaying and recording all the neurons in the entire brain will be feasible in the future. However, resources from funding agencies (e.g., The BRAIN Initiative, the SIMONS foundation, and others), large-scale research projects (such as the Human Brain Project, The Connectome Project, and the Brain Activity Map Project), and the barrage of new studies and new technologies mentioned above [22,23] show the implicit (if not explicit) steps toward this goal. As the community has decided to step in this direction and prioritize the development of “large-scale monitoring” and “precise interventional tools” (BRAIN Initiative recommendations for 2025), we should consider the consequences of this endeavor for solving the fundamental problem of consciousness if/when it is successful.

Our immediate aim is to challenge the primacy of action potentials as an explanation for consciousness. Action potentials are the brain’s main signaling mechanisms, and they form the basis for neural computation as we understand it. But our broader goal is to clarify the limitations of measured neural biological and electrical properties in laboratory settings to explain consciousness.

An overview and discussion of the experimental steps Initially, we recorded the neuronal firing in the entire brain of a participant while she was consciously experiencing green light. In Step 1, we played the recorded action potentials back to each neuron in the participant’s brain (replay), artificially recreating the brain activity that was naturally caused by the visual stimulus. It is important to note that, although the replay was artificial in our experiment, it is a known biological process thought to underlie perception, memory consolidation, and network homeostasis in animals [49] and humans [50–52]. Because the effect of synaptic connections on brain activity was already precluded in Step 1 by imposing the replay, disconnecting the synapses either optogenetically or physically (in Step 2 and Step 3, respectively) has no further consequences on the firing of action potentials. Therefore, if the reader does not reject the hypothesis at the first step, why reject it in further steps (see further discussion in theories of consciousness)? Finally, we argued that if the participant consciously perceived the green light after the resective surgery (Step 3), it would imply that a scattered brain can be conscious. In the working hypothesis, we only considered whether action potentials cause consciousness. Performing our experiments for other neuronal processes might be more difficult than for action potentials and, in some cases, even impossible. However, conceptually, it is straightforward to include them in the hypothesis and even include combinations of multiple processes; for example, membrane potential fluctuations, calcium ion concentrations [53,54], the release of neurotransmitters from the presynaptic terminals, or activity in glial cells [55,56]. To consider multiple biological processes, we first need to record these processes and then test the hypothesis against Steps 1 to 3 by asking in each step whether the participant’s conscious perception changed when the respective cellular processes remained exactly the same. Bayne and colleagues [57] discuss the circumstances, plausibility, and consequences of “islands of awareness” occurring in brains that cannot interact with the natural world via perception and action. Interestingly, Bayne and colleagues argue that islands of awareness can also exist when the brain is physically isolated from a body. Such cases are ex cranio brains, disconnected hemispheres post-hemispherectomy, and cerebral organoids. It is worth noting that the brain during the replay is fundamentally different from such islands of awareness because it effectively contains both the input and the output. The stimulus triggered the trains of action potentials that constitute the replay. The replayed action potentials activate the spinal motoneurons causing the participant to press the button. Although it is difficult to assess whether the brain is conscious during the replay, it is not isolated from the world.

Implications for theories of consciousness The experiment we described here is useful as a benchmark for theories of consciousness, revealing hidden incoherences and ambiguities [58]. Specifically, for a given theory of consciousness, we ask in which step (i.e., Steps 1 to 3) and why we would reject the working hypothesis and claim that the participant loses consciousness. By our estimation, several theories of consciousness would predict that the participant is conscious after all the manipulations from Step 1 to Step 3. In particular, theories that specify the neurobiological mechanisms for consciousness in detail—unless they make some additional assumptions—are forced to conclude that scattered brains are conscious. If conscious experience is caused by action potentials fired by particular neurons, as in the theory by Crick and Koch [25,59], why should disconnecting these neurons or scattering them affect consciousness? Even considering further details of this theory, e.g., that the brainstem and higher-order thalamic nuclei have an enabling role in consciousness [59,60], the theory does not conflict with a conscious participant throughout Steps 1 to 3. Recurrent processing (RP) theory by Lamme [61,62] relates consciousness to feedback between different cortical areas. The feedback is essentially the influence of some neurons, causing action potentials in other neurons. The firing caused by the feedback could just as well be replayed even in disconnected and scattered scenarios. Again, if the role of feedback is to cause action potentials in certain neurons, then nothing in this theory suggests a loss of consciousness in Steps 1 to 3. We recently proposed the dendritic integration theory [DIT; 63,64], which hypothesizes that consciousness depends on the reintegration of top-down information via the apical dendrites of layer 5 pyramidal neurons. DIT is based on the empirical finding that the electrical coupling between apical and basal dendrites of cortical pyramidal neurons is disrupted by common anesthetics, thus blocking the influence of the apical dendrite on the output of the layer 5 pyramidal neurons [65]. According to this theory, decoupling the apical from the basal dendrites in a sufficiently large number of cortical pyramidal neurons would switch off consciousness. Essentially, DIT pinpoints the nexus of information flow within the brain microarchitecture that is crucial for consciousness. Besides the clinical benefit and understanding of the computation involved, DIT offers a framework for interrogating this biological mechanism in the laboratory. Regarding the thought experiment presented here, however, placing an electrode at the cell body to generate the cellular output effectively bypasses the critical nexus point in the apical dendrite. We, therefore, predict that the replay of activity at the cell bodies of pyramidal cells would, in this case, completely entail the former influence of the apical dendrite. Furthermore, DIT is agnostic about the intrinsic necessity of apical causality, per se, versus the resultant firing activity at the cell body. In this respect, DIT does not inform us whether the brain is conscious under replay or whether scattered brains are conscious. A similar conclusion is implied by functionalist theories, which do not commit to a particular neural implementation, but rather suggest that consciousness is related to specific functions or processes regardless of the exact implementation. For instance, the global neuronal workspace (GNW) theory [66–68] suggests that “global broadcasting constitutes the physiological correlate of conscious reportability” [69]. Similarly, the higher-order thought (HOT) theory of consciousness is a functionalist theory that relates consciousness to metacognitive higher-order processing [70,71]. As “global broadcasting” and higher-order processing are functions that are not restricted to brains [72,73], these theories do not necessarily conflict with the notion that the replay generates consciousness (Step 1). In particular, the same patterns of global broadcasting, self-sustained activity and ignition dynamics central to the GNW [66,68], and the higher-order processing central to HOT [70,71] could be exactly replayed in Step 1. The proponents of these theories might be more skeptical of Step 2 and Step 3, but nothing inherent to these theories would prohibit a disconnected or scattered brain from being conscious. According to Zeki and colleagues [74], consciousness comprises nodes of micro-consciousnesses in different brain regions. Interestingly, in contrast to other theories, Zeki argues that consciousness is not unified [75]. Therefore, color and motion, for example, are consciously perceived in different parts of the cortex and only then bind together with other nodes to form a macro-consciousness. A micro-consciousness is autonomous [76] and does not require further processing. Therefore, the resected visual cortex in Step 3 may become micro-conscious of the green light during the replay. We could not find a direct reason as to why, according to Zeki, scattered brains during replay cannot bind together into a macro-consciousness. In conclusion, some theories do not have conceptual reasons or assumptions as to why replaying, disconnecting, or scattering the neurons should lead to a loss of consciousness. What are the additional assumptions needed to escape these conclusions? Essentially there are 2 possibilities: one, that the three-dimensional structure of the brain is necessary for consciousness, and the other that the intrinsic cause and effect (i.e., between the neurons) is necessary (see the “The replay conundrum” section). Even if these theories make one of these additional assumptions, what is the justification? Is it only an ad hoc assumption to avoid the conclusion that scattered brains could be conscious? Consciousness is lost in Step 3: Theories that require the structure of the brain. The brain’s particular three-dimensional structure is central for theories that associate consciousness with the electromagnetic field [77–83]. These theories would predict that the participant consciously experiences green light as long as the electrical field remains as it was during the recording. Our participant, therefore, will remain conscious of the green light during the replay (Step 1) and after synaptic disconnection (Step 2), given that these steps do not interfere with the brain’s field relevant to consciousness. However, the participant’s conscious experience may change by surgically removing the visual cortex (Step 3), altering the brain’s physical structure and, consequently, the electromagnetic field. This offers an explanation as to why a scattered brain cannot be conscious. Consciousness is lost in Step 1: Theories that require intrinsic cause and effect. Integrated information theory [84–86] quantifies consciousness based on the repertoire of all possible cause-and-effect interactions between the neurons in the brain’s network. Disconnecting the neurons in Step 2 abolished the network structure that underlies the interaction between neurons. However, in Step 1, the replay imposed particular (recorded) trains of action potentials and effectively vetoed all the interaction between the neurons, even though the synaptic connections were fully functional. Therefore, according to the assumptions of IIT, our participant already loses consciousness in Step 1. According to Searle’s “biological naturalism” [87] (which is an approach rather than a theory that does not specify the biological mechanism for consciousness), the replay in Step 1 does not necessarily result in loss of consciousness. The participant will remain conscious during the replay as long as the underlying biological substrate and the “right” causal powers are intact. Therefore, according to Searle, it is not evident if and in which precise step loss of consciousness would happen. The right causal powers may lie, for example, in the propagation of the action potentials within the axon. In this case, our participant would remain conscious during the replay because both the naturally occurring and the artificially induced action potentials propagate via the axon. According to Searle, replaying other biological processes may have different outcomes. Therefore, the replay could be insightful in revealing the particular causal powers that matter for consciousness.

The replay conundrum To underscore the usefulness of replay as a potential experimental strategy, let us compare the replay of brain activity to a detailed simulation of the brain. A frequent objection to the view that a detailed simulation of the human brain can become conscious is that it merely manipulates symbols whose meaning depends on external interpretation, whereas neural activity is intrinsically meaningful to the brain [88]. In contrast to a simulation, the artificial neuronal firing induced by the replay is intrinsically meaningful to the brain/participant because it is an identical copy of intrinsically meaningful activity (i.e., an experience of green light). John Searle famously explained that “you could not digest pizza by running the program that simulates such digestion” [89]. Unlike biologically detailed simulations running on a computer, the replay is recorded and activated on the same substrate. Therefore, in contrast to a simulation of the stomach, recording and then replaying smooth muscle contraction and enzyme secretion would result in digestion. What would it imply about the nature of consciousness if replay would work for stomach digestion or the heart pumping blood but not for the brain and consciousness? Several theories of consciousness claim that cause-and-effect relations among the neurons within the brain are decisive for consciousness, either with or without strictly insisting on the necessity of the biological substrate. For some theories, cause and effect between the neurons is no more than a mechanism responsible for generating patterns of brain activity. For other theories, cause and effect is more fundamental, and without it, consciousness cannot exist (for extended discussion, see the previous section). Because the replay abolishes the biological intrinsic cause-and-effect relations between the neurons, the later theories assert that our participant loses consciousness during the replay. However, the analogous assertion concerning the heart and stomach does not work; even without cause and effect between the elements of the heart, replaying myocardial cells’ activity would do the biological and mechanical work of pumping blood. Why should a replay of action potentials in the case of the brain and consciousness not have a similar outcome to the case of stomach and food digestion or heart and pumping blood? Possibly, cause and effect between the neurons is required only in the case of the brain and consciousness. The replay could be viewed as analogous to brain lesion experiments, but rather than removing a piece of tissue, we remove cause and effect to examine whether the basis for consciousness is either neuronal activity or cause and effect (or both). Our experiment is a plausible and decisive strategy to distinguish between these 2 possibilities.

[END]

[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001651

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.