Jean-Dominique Bauby, who once served as the editor-in-chief of the French fashion magazine Elle, describes in his book The Diving Bell and the Butterfly the sensation of a patient with locked-in syndrome being imprisoned in an immobile body, based on his own experiences.

Before this, Alexandre Dumas had already depicted a scenario in The Count of Monte Cristo where a fully intact consciousness is buried within an immobile body—Monsieur Noirtier de Villefort cannot speak or move his limbs, but through eye movements, he tries to prevent a murder and an undesirable marriage.

These real or fictional cases reflect a widespread concern about the “distribution of consciousness.”

▷ The Diving Bell and the Butterfly film adaptation

In fact, “consciousness” is a concept that has only flourished in recent times. Although research related to consciousness can be traced back to ancient Greece, and even hints of consciousness and the subconscious can be found in primitive cave paintings, research focusing specifically on consciousness itself has a very short history.

In the early 1980s, the term “consciousness” was still taboo in serious scientific publications. Many researchers believed that defining consciousness was outdated and ambiguous, and that the use of the term “consciousness” added no value to psychology.

It was not until the late 1980s that consciousness research took a turn for the better*. The intuitive ideas about the “distribution of consciousness” in folk psychology have also received increasing attention [1]. We usually think that people in a waking state are conscious, while in states of intoxication, mental illness, anesthesia, coma, vegetative state, and brain death, the degree of consciousness progressively decreases.

For research on the shift in attitudes toward consciousness, refer to Stanislas Dehaene’s Consciousness and the Brain. This observation is partly based on Dehaene’s personal experience and partly on his academic research.

In addition to folk psychology, clinical practice also pays significant attention to the issue of the distribution of consciousness. Clinically, diagnosing whether a patient is brain-dead or in a vegetative state requires measuring the degree of consciousness. Furthermore, from a broader perspective, how do we determine the existence of consciousness? Which animals can be included in the “consciousness club,” and in what ways do they possess consciousness? For example, is the consciousness of an octopus dispersed or unified? Do organelles or artificial intelligence systems, such as large language models with strong communicative abilities, possess consciousness? These studies from different perspectives and levels ultimately point to the investigation of the nature of consciousness—is consciousness unified? Is consciousness multiply realizable (e.g., through physical, computational, or modeling means)?

Currently, there are roughly 22 theories of consciousness, and to adjudicate the distribution of consciousness, we need to base our judgments on a universally accepted theory of consciousness. However, most theories of consciousness are centered on human consciousness, making it very difficult to explain the distribution of consciousness in animals.

To address this, Tim Bayne, Anil Seth, and others have taken a different approach. In the paper “Tests for Consciousness in Humans and Beyond,” they attempt to provide a framework for consciousness tests based on the key characteristics of consciousness testing and offer strategies for proving the validity of these tests. This builds a bridge between consciousness assessments and theoretical frameworks, ultimately leading to a deeper understanding of the nature of consciousness. This testing framework can not only be used to evaluate and revise existing consciousness tests but also guide the construction of new ones.

▷ Bayne, Tim, et al. “Tests for consciousness in humans and beyond.” Trends in Cognitive Sciences (2024).

1. From the Hard Problem to the Real Problem

In Consciousness and the Brain, Stanislas Dehaene identifies three key elements that have revitalized consciousness research: a more precise definition of consciousness; discoveries that allow for the experimental manipulation of consciousness; and the academic community’s renewed emphasis on studying subjective phenomena. These factors have collectively helped consciousness research emerge from its “winter.” The advent of new research tools such as electroencephalography (EEG), functional magnetic resonance imaging (fMRI), and magnetoencephalography (MEG) has accelerated this progress.

The three elements emphasized by Dehaene are directly related to the core features of consciousness research. Among them, “experimentally manipulable consciousness” primarily focuses on the accessibility of consciousness and first-person reports. “Subjective phenomena” refer to the “what it is like” subjective characteristics highlighted by Thomas Nagel in his essay “What Is It Like to Be a Bat?” or what Ned Block describes as phenomenal consciousness.

Additionally, the characteristics pointed out by Dehaene reveal the challenges in consciousness research, specifically the distinction between the “hard problem” and the “easy problem” of consciousness as articulated by David Chalmers. The hard problem concerns how phenomenal consciousness arises from its physical basis and how to account for the reality of phenomenal consciousness in the universe. In contrast, the easy problem addresses how physical systems generate responses with specific functional or behavioral performances. Simply put, the hard problem deals with how a physical brain can possess phenomenality, while the easy problem examines how particular types of conscious experiences are produced.

Although addressing the easy problem is not without difficulty, it is relatively straightforward because there is no conceptual gap between physical systems and their functional or behavioral manifestations. However, even if all easy problems are solved, the hard problem remains unresolved. As George Mather stated, the “hardness” that Chalmers refers to implies that solving this problem seems impossible.

Subsequently, the philosophy of mind and consciousness science have pursued the mysteries of consciousness along different paths. The philosophy of mind has developed a series of theories, such as physicalism and functionalism, in an attempt to provide answers to the hard problem. Consciousness science, on the other hand, avoids direct discussion of phenomenality and approaches the subject through first-person reports and the accessibility of consciousness.

For example, the Global Workspace Theory and Higher-Order Thought Theory focus on the function or behavioral manifestations of consciousness. The former posits that the core of consciousness lies in the accessibility of information within a global workspace, while the latter emphasizes the reference of higher-order representations to first-order representations, reflected in the brain as the prefrontal cortex referencing other brain regions. Even if subjective reports of conscious phenomena largely align with the explanations provided by these theories, this consistency only indicates some form of correlation and does not directly explain the phenomena themselves.

▷ For Global Neural Workspace Theory, see: Is a Grand Unified Theory of Consciousness Coming?

Anil Seth points out that we should neither be confined to debates over the hard problem nor neglect the phenomenality of consciousness. Instead, we should shift our attention to the real problem of consciousness. The real problem of consciousness attempts to integrate phenomenality, measurement, and explanations of consciousness to explain, predict, and control phenomenally characterized experiences. Anil Seth believes that we can anticipate that answers to the real problem will ultimately allow the hard problem to “fade into the metaphysical fog.” This shift in perspective provides us with a new path for studying consciousness. The transition from the hard problem to the real problem places consciousness testing and theoretical frameworks in a more central role, with the two mutually clarifying and corroborating each other, thereby contributing to a deeper understanding of consciousness.

2. Measurement of Consciousness

When discussing consciousness, we often consider related concepts such as wakefulness, attention, intelligence, voluntary behavior, and self-regulation. Generally speaking, individuals in a state of wakefulness are deemed conscious. Vegetative patients, while still experiencing day-night cycles, do not clearly possess consciousness. The emergence of conscious mental states requires the operation of attention. However, even sub-threshold operations of attention contribute to the manifestation of mental states. Therefore, attention is not synonymous with consciousness. Bayne et al. point out that consciousness tests should not focus solely on the abilities associated with consciousness mentioned above but should instead target phenomenal consciousness directly. Abilities that covary with consciousness, such as attention and perceptual organization, are at best tools for guiding the search.

The scope of consciousness research cannot be too broad or too narrow. Humans, capuchin monkeys, mice, and octopuses exhibit different manifestations of consciousness; these different manifestations are neither modeled on human consciousness nor derived from it. In fact, human consciousness represents just a small part of the broader landscape of consciousness. Therefore, defining the boundaries of consciousness research must consider these diverse manifestations, which in turn require distinct tests.

▷ Examples of Consciousness Tests.

3. Standard Consciousness Tests

The term “consciousness” typically refers to a state of consciousness, but this does not presuppose the existence of a universally applicable and definitive consciousness test. Methods for assessing consciousness are diverse, each with its own focus. Some consciousness tests concentrate on the general features of consciousness without conveying information about its content, such as the Perturbational Complexity Index (PCI) test. These types of tests focus on neural integration and differentiation rather than on individual behavioral responses or reactions to mental imagery.

Another category of consciousness tests focuses on specific conscious contents or mental abilities sufficient to trigger consciousness. These include bodily sensations (such as pain and smell), autonomous responses (generating and maintaining mental imagery), the ability to discriminate between sub-threshold and supra-threshold stimuli, and various types of learning abilities. The interpretation of these tests depends on whether the selected mental abilities are strong indicators of consciousness.

Faced with such diversity, how can we determine whether a particular consciousness test is applicable to a specific group? Why can the results of a consciousness test be regarded as reliable evidence of the presence or absence of consciousness? Additionally, how should we understand the relationships between different types of consciousness tests? To address these questions, Bayne et al. propose a “four-dimensional space of consciousness tests,” aiming to provide a more systematic framework for understanding consciousness tests.

4. The Four-Dimensional Space of Consciousness Tests

The first dimension of the four-dimensional space emphasizes the limited applicability of consciousness tests. In other words, different consciousness tests should be designed for target groups exhibiting varying manifestations of consciousness. Some consciousness tests may be suitable only for humans and other primates; others might apply to a broader range of mammals, and still, others could be utilized for more diverse entities, including evolving biological systems and artificial intelligence. Ideally, a universally applicable consciousness test would encompass all types, but this is unlikely to be achievable in the short term.

Furthermore, the goal of consciousness tests is to accurately determine whether an individual possesses consciousness. This means that the tests should neither incorrectly classify unconscious individuals as conscious (false positives) nor mistakenly identify conscious individuals as unconscious (false negatives). The former is referred to as the specificity of the consciousness test, measured by a low false positive rate; the latter is known as sensitivity, measured by a low false negative rate. These two concepts—specificity and sensitivity—constitute the remaining two dimensions of our discussed four-dimensional space.

▷ Example: Sensitivity refers to the proportion of truly unconscious individuals who are correctly diagnosed as unconscious. Specificity refers to the proportion of truly conscious individuals who are correctly diagnosed as conscious. More specifically, false positives refer to individuals who have been incorrectly diagnosed as unconscious; False negative refers to a person who is mistakenly diagnosed as conscious.

Take, for example, the command-following test for human consciousness. This test is commonly used with patients who exhibit no observable behavioral responses. By monitoring neural activity in response to mental commands, we can assess whether they possess consciousness. In many clinical cases, this test is considered a reliable indicator of the presence of consciousness and is also a crucial tool for determining minimal consciousness states.

However, there are exceptions to this test. Some epilepsy patients may pass the command-following test, but their responses might merely be unconscious habitual actions, such as automatic walking, which do not conclusively demonstrate consciousness. Additionally, conscious patients might fail the test if they do not hear or understand the commands.

This implies that even consciousness tests are not always entirely accurate. The specificity (the ability to correctly exclude unconsciousness) and sensitivity (the ability to correctly identify consciousness) of a test do not always align. A test with high specificity but low sensitivity might miss genuinely conscious patients, while a test with high sensitivity but low specificity might incorrectly classify unconscious patients as conscious.

Specificity and sensitivity are based on the limited applicability of consciousness tests and depend on the specific target group. We cannot expect high specificity for a particular group to transfer to another type of target group. Low-specificity consciousness tests are not entirely useless; they may indicate that current research on the distribution of consciousness within specific groups is misguided.

As previously mentioned, the false positive rate and false negative rate are important indicators of specificity and sensitivity, respectively. However, these are statistical terms that can only provide inductive explanations. Therefore, the final dimension of consciousness tests is rational confidence. This is a second-order criterion that assesses whether the specificity and sensitivity are reasonable, attempting to link scientific tests with intuitive judgments about consciousness in folk psychology*. Tests with high rational confidence better align with our everyday understanding of consciousness, whereas tests with low rational confidence may yield results that starkly contrast with our intuitions, making their outcomes more suggestive than definitive. Establishing this dimension essentially measures the overall efficacy of a consciousness test.

Achieving high rational confidence is challenging in the short term, especially for consciousness tests designed to assess the consciousness levels of other animals, given our limited folk psychological judgments about the distribution of consciousness in non-human species.

▷ Scope of Consciousness Tests: The applicability of different consciousness tests (rows) to various target groups (columns). A plus sign (+) indicates that the test may meaningfully apply to a specific population (although its specificity/sensitivity might be low) and may require some modifications. A dash (−) signifies that the test is not applicable or irrelevant to a particular population. A question mark (?) suggests that the test might be applicable but requires further development to determine its suitability. Lastly, a combination of a plus sign and a question mark (+?) indicates that while the test can be applied, it is unclear what the results signify.

5. How to Prove the Validity of Consciousness Tests

How can we demonstrate the validity of a consciousness test and assess its rational confidence? How can we determine the specificity and sensitivity of consciousness tests? First and foremost, we cannot simply conclude that a consciousness test is valid by proving that a particular group possesses consciousness. Instead, we must first establish that the consciousness test is usable and that its results are reliable before we can determine the degree of consciousness present in the target group. Therefore, demonstrating the validity of consciousness tests is crucial.

In this context, Seth, Bayne, and others outline three strategies for proving the validity of consciousness tests in their article. These strategies are not mutually exclusive; they merely focus on different core factors. The redeployment strategy assumes the validity of certain consciousness tests and uses this assumption as a foundation to discuss the validity of other related tests. However, the validity of the foundational tests themselves remains to be demonstrated. The theory-based strategy offers a more robust foundation compared to the redeployment strategy. However, since the field of consciousness theories has not yet reached a definitive conclusion and different consciousness theories support different consciousness tests, there is no unified answer as to which theory should serve as the basis for consciousness tests. Compared to the first two, the iterative natural kind strategy may be a better choice. This strategy views consciousness as a natural kind, providing a basis for further reasoning and generalization. Its iterative nature creates a positive feedback loop between consciousness tests and consciousness theories.

Strategy One: The Redeployment Strategy

The redeployment strategy posits that there are already some consciousness tests with broad validity, and other tests to be considered are variants of these existing tests. Therefore, we can extend the validity of existing tests to their variants. In everyday life, we generally acknowledge that overt behavior signifies consciousness, and covert maintenance of mental imagery is a variant of overt behavior. The command-following test uses this as an indicator to provide evidence for the presence or absence of consciousness.

This is a relatively conservative strategy, merely extending existing valid consciousness tests. However, it also involves a risky leap. Firstly, this strategy is entirely empirical, seemingly assuming that existing consciousness tests have sufficiently validated their effectiveness. However, their validity is only empirically grounded and does not constitute rational justification. Secondly, the extension has boundaries; this strategy only allows for the extension of the original test’s content variants and, in principle, cannot be applied to other target groups. Thus, it ultimately only tests the same target group’s consciousness state in different ways. For example, using a variant of the command-following test to measure the consciousness state of human patients does not allow this test to be applied to artificial intelligence systems.

For the redeployment strategy to achieve a robust foundation, it requires proving the validity of the foundational tests through other means or accepting the deflationism of consciousness, which holds that any system that can reliably follow commands is conscious. However, we cannot prove the validity of consciousness independently of consciousness tests, nor do we have a prior definition of consciousness. Both paths are unfeasible.

▷ Katya Dorokhina

Strategy Two: The Theory-Based Strategy

The redeployment strategy either seeks another foundation or embraces some form of deflationism of consciousness. In any case, this strategy faces fundamental challenges. Since the redeployment strategy lacks a theoretical foundation, some have proposed invoking consciousness theories to demonstrate the validity of consciousness tests. The basic premise of this strategy is that consciousness tests that fit well with well-grounded and validated consciousness theories can serve as reliable indicators of consciousness. For example, the Global Workspace Theory supports the Global Effect Test, and the Information Integration Theory inspires the Perturbational Complexity Index (PCI) test.

The theory-based strategy also encounters several challenges. As previously mentioned, no single consciousness theory has gained widespread acceptance; at least 22 consciousness theories are being seriously considered, some of which have variants, and these theories support different consciousness tests*. If these diverse theories could eventually be integrated, the existing disagreements would no longer pose a challenge. However, this is not the case, as different consciousness theories have not shown a trend towards convergence.

There are instances where different theories support the same consciousness test. Farisco and Changeux, in their 2023 article, discuss the fundamental compatibility between the Perturbational Complexity Index test and the Global Neural Workspace Theory. Nonetheless, overall, different consciousness theories support different consciousness tests.

Based on this, Chalmers and others suggest that an integrated framework for consciousness theories might resolve this issue, with different consciousness theories holding varying weights within this framework and multiple consciousness tests being combined accordingly. This idea may be feasible, but in practice, reaching a consensus on the weights of consciousness theories is challenging, much like the unresolved disputes within consciousness theories themselves.

This strategy may also face the challenge of anthropocentrism. Many consciousness theories are built with humans as the reference point, raising questions about how these theories apply to other groups. For instance, the Global Workspace Theory explains the mechanism of human consciousness but does not specify which systems possess a global workspace. There is evidence that fish have structures similar to a global workspace, and the lateral neurons in birds’ tails might perform functions analogous to the human prefrontal cortex, but how to determine their similarity remains unclear. Beyond these animals, how should other animals be treated?

The third challenge involves the bidirectional dependency between consciousness science and consciousness philosophy. To validate a consciousness theory, one needs to rely on consciousness tests; simultaneously, consciousness tests require a specific consciousness theory for support. This situation creates a cycle where a theory depends on tests for validation while also being used to validate the tests, forming an intractable loop.

Strategy Three: The Iterative Natural Kind Strategy

To break free from the non-virtuous cycle between consciousness tests and consciousness theories, Bayne and others propose the iterative natural kind strategy. This strategy views consciousness as a natural kind, where natural kinds are categorized based on essential commonalities (rather than superficial similarities, arbitrary feature combinations, or purely human interests), allowing for the division of the world “at its joints” in a way that reflects the structure of the natural world.

▷ Illustration of the Iterative Natural Kind Strategy

If consciousness is a natural kind, this offers several advantages for research. Firstly, different manifestations of consciousness belonging to the same natural kind share the same essence, which can be discovered through iterative steps, much like scientists have uncovered the essence of heat. The process of revealing the essence of heat began with some pre-theoretical assumptions, which were continually refined through the theory’s unity, simplicity, and explanatory power.

The process of uncovering the essence of consciousness could also result from the iteration between experiments and theories, not only aiding in the construction of more accurate consciousness tests but also helping us better understand the distribution of consciousness. Iteration simultaneously involves retaining and transcending; consciousness theories retain some ideas from folk psychology but are not derived from the initial pre-theoretical assumptions. Instead, they transcend and refine these assumptions. The resulting consciousness theories do not contradict folk psychology but are more rigorous, systematic, and grounded in ongoing empirical evidence.

The natural kind strategy can also effectively address the generalization problem. Since consciousness is a natural kind with a shared essence, starting from the characteristics of some groups, we can first extend to proximate groups with similar features and then to other groups with different features. Theoretically, this is feasible, though during the extension process, the confidence in pre-theoretical assumptions diminishes. Therefore, the validity provided by the natural kind strategy is hierarchical.

However, it remains undecided which features to use for expansion and which groups are closest to human consciousness. Different measurement standards can yield different proximities, whether behavioral, functional, or neurophysiological. For example, individuals with a nervous system exhibit varying degrees of response in states of complete motor inhibition, a phenomenon particularly evident in human clinical conditions, human infants, and some animals. In contrast, artificial intelligence systems and cellular structures differ significantly from humans in this regard. Therefore, the better we understand the underlying mechanisms of human consciousness, the more accurately we can stratify different groups.

Since the consciousness of different groups is interrelated, the validity of any consciousness test must rely on other forms of verification. Simultaneously, developing independent and effective consciousness tests is crucial, enabling consciousness tests to not only corroborate each other but also correct one another.

The iterative natural kind strategy presents a virtuous cycle between consciousness tests and consciousness theories, offering a hierarchical and expandable structure for understanding consciousness. This provides a reliable path for deepening our comprehension of consciousness.

6. Conclusion

This paper delineates a four-dimensional space for developing robust and generalizable consciousness tests by capturing the key characteristics of consciousness testing. This framework not only enables us to evaluate the strengths and weaknesses of individual tests and better comprehend how they interrelate but also provides a solid foundation for developing more refined assessments. An ideal consciousness test possesses universal applicability, perfect specificity and sensitivity, and high rational confidence, thereby establishing it as an effective measure of consciousness.

Furthermore, the paper introduces three strategies for validating the effectiveness of these tests: the redeployment strategy, the theory-based strategy, and the iterative natural kind strategy. Among these, the iterative natural kind strategy is considered the ideal choice. This strategy posits that consciousness is a natural kind, and we should begin with tests for groups closely related to us and then systematically and purposefully expand to other groups.

The four-dimensional space and the validation strategies serve as a bridge between consciousness tests and consciousness theories, as well as a nexus connecting third-person experimental results with the subjective characteristics of conscious experience. All of this ultimately aids in addressing critical questions in the fields of consciousness science and philosophy of mind: What is consciousness? Is a unified theory of consciousness possible? How should we understand the relationship between consciousness and folk psychology?

1.Artificial Intelligence vs. Human Intelligence

1.1 How did early artificial intelligence models draw inspiration from our understanding of the brain?

The early development of artificial intelligence was greatly influenced by our understanding of the human brain. In the mid-20th century, advancements in neuroscience and initial insights into brain function led scientists to attempt applying these biological concepts to the development of machine intelligence.

In 1943, neurophysiologist Warren McCulloch and mathematician Walter Pitts introduced the “McCulloch-Pitts neuron model,” one of the earliest attempts in this area. This model described the activity of neurons using mathematical logic, which, although simplistic, has laid the groundwork for later artificial neural networks.

▷ Fig.1: Structure of a neuron and the McCulloch-Pitts neuron model.

During this period, brain research primarily focused on how neurons process information and interact within complex networks via electrical signals. These studies inspired early AI researchers to design primitive artificial neural networks.

In the 1950s, the perceptron, invented by Frank Rosenblatt, was an algorithm inspired by biological visual systems, simulating how the retina processes information by receiving light. Although it was rudimentary, it marked a significant step forward in the field of machine learning.

▷ Fig.2: Left: Rosenblatt’s physical perceptron; Right: Structure of the perceptron system.

In addition to the influence of neuroscience, early cognitive psychology research also contributed to the development of AI. Cognitive psychologists sought to understand how humans perceive, remember, think, and solve problems, providing a methodological foundation for simulating human intelligent behavior in AI. For instance, the logic theorist developed by Allen Newell and Herbert A. Simon could prove mathematical theorems [1-3], simulating the human problem-solving process and, to some extent, mimicking the logical reasoning involved in human thought.

Although these early models were simple, their development and design were profoundly shaped by contemporary understandings of the brain, which established a theoretical and practical foundation for the development of more complex systems. Through such explorations, scientists gradually built intelligent systems capable of mimicking or even surpassing human performance in specific tasks, driving the evolution and innovation of artificial intelligence technology.

1.2 Development of Artificial Intelligence

Since then, the field of artificial intelligence has experienced cycles of “winters” and “revivals.” In the 1970s and 1980s, improvements in computational power and innovations in algorithms, such as the introduction of the backpropagation algorithm, made it possible to train deeper neural networks. During this period, although artificial intelligence achieved commercial success in certain areas, such as expert systems, limitations that arose from technology and overly high expectations ultimately led to the first AI winter.

Entering the 21st century, especially after 2010, the field of artificial intelligence has witnessed unprecedented advancements. The exponential growth of data, the proliferation of high-performance computing resources like GPUs, and further optimization of algorithms propelled deep learning technologies as the main driving force behind AI development.

The core of deep learning remains the simulation of how brain neurons process information, and its applications have far surpassed initial expectations, encompassing numerous fields such as image recognition, natural language processing, autonomous vehicles, and medical diagnostics. These groundbreaking advancements have not only driven technological progress but also fostered the emergence of new business models and rapid industry development.

▷ Giordano Poloni

1.3 Differences Between Artificial Intelligence and Human Intelligence

1.3.1 Differences in Functional Performance

Although artificial intelligence has surpassed human capabilities in specific domains such as board games and certain image and speech recognition tasks, it generally lacks cross-domain adaptability. While some AI systems, particularly deep learning models, excel in big data environments, they typically require vast amounts of labeled data for training, and their transfer learning abilities are limited when tasks or environments change, often necessitating the design of specific algorithms. In contrast, the human brain possesses a robust learning and adaptation capacity, which is able to learn new tasks with minimal data across various conditions and perform transfer learning, applying knowledge gained in one domain to seemingly unrelated areas.

In terms of flexibility in addressing complex problems, AI performs best with well-defined and structured issues, such as board games and language translation. However, its efficiency drops when dealing with ambiguous and unstructured problems, which makes it susceptible to interference. The human brain exhibited high flexibility and efficiency in processing vague and complex environmental information; for instance, it can recognize sounds in noisy environments and make decisions despite incomplete information.

Regarding consciousness and cognition, current AI systems lack true awareness and emotions. Their “decisions” are purely algorithmic outputs based on data, devoid of subjective experience or emotional involvement. Humans, on the other hand, not only process information but also possess consciousness, emotions, and subjective experiences, which are essential components of human intelligence.

In multitasking, while some AI systems can handle multiple tasks simultaneously, this often requires complex, targeted designs. Most AI systems are typically designed for single tasks, and their efficiency and effectiveness in multitasking typically do not match those of the human brain, which can quickly switch between tasks while maintaining high efficiency.

In terms of energy consumption and efficiency, advanced AI systems, especially large machine learning models, often demand significant computational resources and energy, far exceeding that of the human brain. The brain operates on about 20 watts, showcasing exceptionally high information processing efficiency.

Overall, while artificial intelligence has demonstrated remarkable performance in specific areas, it still cannot fully replicate the human brain, particularly in flexibility, learning efficiency, and multitasking. Future AI research may continue to narrow these gaps, but the complexity and efficiency of the human brain remain benchmarks that are difficult to surpass.

▷ Spooky Pooka ltd

1.3.2 Differences in Underlying Mechanisms

In terms of structure, modern AI systems, especially neural networks, are inspired by biological neural networks, yet the “neurons” (typically computational units) and their interconnections rely on numerical simulations. The connections and processing in these artificial neural networks are usually pre-set and static, lacking the dynamic plasticity of biological neural networks. The human brain comprises approximately 86 billion neurons, each connected to thousands to tens of thousands of other neurons via synapses [6-8], supporting complex parallel processing and highly dynamic information exchange.

Regarding signal transmission, AI systems transmit signals through numerical calculations. For instance, in neural networks, the output of a neuron is a function of the weighted sum of its inputs, processed using simple mathematical functions such as Sigmoid or ReLU. Neural signal transmission relies on electrochemical processes, where information exchange between neurons occurs through the release of neurotransmitters at synapses, regulated by various biochemical processes.

In terms of learning mechanisms, AI learning typically adjusts parameters (such as weights) through algorithms, such as backpropagation. Although this method is technically effective, it requires substantial amounts of data and necessitates retraining or significant adjustment of model parameters for new datasets, highlighting a gap compared to the brain’s continuous and unsupervised learning approach. Learning in the human brain relies on synaptic plasticity, where the strength of neural connections changes based on experience and activity, supporting ongoing learning and memory formation.

1.4 Background and Definition of the Long-Term Goal of Simulating Human Intelligence—Artificial General Intelligence

The concept of Artificial General Intelligence (AGI) arose from recognizing the limitations of narrow artificial intelligence (AI). Narrow AI typically focuses on solving specific, well-defined problems, such as board games or language translation, but lacks the flexibility to adapt across tasks and domains. As technology advances and our understanding of human intelligence deepens, scientists begin to envision an intelligent system with human-like cognitive abilities, self-awareness, creativity, and logical reasoning across multiple domains.

AGI aims to create an intelligent system capable of understanding and solving problems across various fields, with the ability to learn and adapt independently. This system would not merely serve as a tool; instead, it would participate as an intelligent entity in human socio-economic and cultural activities. The proposal of AGI represents the ideal state of AI development, aspiring to achieve and surpass human intelligence in comprehensiveness and flexibility.

2. Pathways to Achieving Artificial General Intelligence

Diverse neuron simulations and network structures exhibit varying levels of complexity. Neurons with richer dynamic descriptions possess higher internal complexity, while networks with wider and deeper connections exhibit greater external complexity. From the perspective of complexity, there are currently two promising pathways to achieve Artificial General Intelligence: one is the external complexity large model approach, which involves increasing the width and depth of the model; the other is the internal complexity small model approach, which entails adding ion channels to the model or transforming it into a multi-compartment model.

▷ Fig.3: Internal and external complexity of neurons and networks.

2.1 External Complexity Large Model Approach

In the field of artificial intelligence (AI), researchers increasingly rely on the development of large AI models to tackle broader and more complex problems. These models typically feature deeper, larger, and wider network structures, known as the “external complexity large model approach.” The core of this method lies in enhancing the model’s ability to process information (especially when dealing with large data sets) and learn by scaling up the model.

2.1.1 Applications of Large Language Models

Large language models, such as GPT (Generative Pre-trained Transformer) and BERT (Bidirectional Encoder Representations from Transformers), are currently hot topics in AI research. These models learn from vast text data using deep neural networks, master the deep semantics and structures of language, and demonstrate exceptional performance in various language processing tasks. For instance, GPT-3, trained on a massive text dataset, not only generates high-quality text but also performs tasks like question answering, summarization, and translation.

The primary applications of these large language models include natural language understanding, text generation, and sentiment analysis, making them widely applicable in fields such as search engines, social media analysis, and automated customer service.

2.1.2 Why Expand Model Scale?

According to research by Jason Wei, Yi Tay, William Fedus, and others in “Emergent Abilities of Large Language Models,” as the model size increases, a phenomenon of “emergence” occurs, where certain previously latent capabilities suddenly become apparent. This is due to the model’s ability to learn deeper patterns and associations by processing more complex and diverse information.

For example, ultra-large language models can exhibit problem-solving capabilities for complex reasoning tasks and creative writing without specific targeted training. This phenomenon of “emergent intelligence” suggests that by increasing model size, a broader cognitive and processing capability closer to human intelligence can be achieved.

▷ Fig.4: The emergence of large language models.

2.1.3 Challenges

Despite the unprecedented capabilities brought by large models, they face significant challenges, particularly concerning efficiency and cost.

First, these models require substantial computational resources, including high-performance GPUs and extensive storage, which directly increases research and deployment costs. Second, the energy consumption of large models is increasingly concerning, affecting their sustainability and raising environmental issues. Additionally, training these models requires vast amounts of data input, which may lead to data privacy and security issues, especially when sensitive or personal information is involved. Finally, the complexity and opacity of large models may render their decision-making processes difficult to interpret, which could pose serious problems in fields like healthcare and law, where high transparency and interpretability are crucial.

2.2 Internal Complexity Small Model Approach

When discussing large language models, they leave a strong impression with their highly “human-like” output capabilities. Webb et al. examined the analogical reasoning abilities of ChatGPT [3] and found that it has emerged with zero-shot reasoning capabilities, enabling it to address a wide range of analogical reasoning problems without explicit training. Some believe that, if large language models (LLMs) like ChatGPT can indeed produce human-like responses to common psychological measures (such as judgments about actions, recognition of values, and views on social issues), they may eventually replace human subject groups in the future.

2.2.1 Theoretical Foundations

Neurons are the fundamental structural and functional units of the nervous system, primarily composed of cell bodies, axons, dendrites, and synapses. These components work together to receive, integrate, and transmit information. The following sections will introduce the theoretical foundations of neuron simulation, covering neuron models, the conduction of electrical signals in neuronal processes (dendrites and axons), synapses and synaptic plasticity models, and models with complex dendrites and ion channels.

▷ Fig.5: Structure of a neuron.

(1) Neuron Models

Ion Channels

Ion channels and pumps in neurons are crucial membrane proteins that regulate the transmission of neural electrical signals. They control the movement of ions across the cell membrane, thereby influencing the electrical activity and signal transmission of neurons. These structures ensure that neurons can maintain resting potentials and generate action potentials, forming the foundation of the nervous system’s function.

Ion channels are protein channels embedded in the cell membrane that regulate the passage of specific ions (such as sodium, potassium, calcium, and chloride). Various factors, including voltage changes, chemical signals, and mechanical stress, control the opening and closing of these ion channels, impacting the electrical activity of neurons.

▷ Fig.6: Ion channels and ion pumps for neurons.

Equivalent Circuit

The equivalent circuit model simulates the electrophysiological properties of neuronal membranes using circuit components, allowing complex biological electrical phenomena to be explained and analyzed within the framework of physics and engineering. This model typically includes three basic components: membrane capacitance, membrane resistance, and a power source.

The cell membrane of a neuron exhibits capacitive properties related to its phospholipid bilayer structure. The hydrophobic core of the lipid bilayer prevents the free passage of ions, resulting in high electrical insulation of the cell membrane. When the ion concentrations differ on either side of the cell membrane, especially under the regulation of ion pumps, charge separation occurs. Due to the insulating properties of the cell membrane, this charge separation creates an electric field that allows the membrane to store charge.

Capacitance elements are used to simulate this charge storage capability, with capacitance values depending on the membrane’s area and thickness. Membrane resistance is primarily regulated by the opening and closing of ion channels, directly affecting the rate of change of membrane potential and the cell’s response to current input. The power source represents the electrochemical potential difference caused by the ion concentration gradient across the membrane, which drives the maintenance of resting potential and the changes in action potential.

▷ Fig.7: Schematic diagram of the equivalent circuit.

Hodgkin-Huxley Model

Based on the idea of equivalent circuits, Alan Hodgkin and Andrew Huxley proposed the Hodgkin-Huxley (HH) model in the 1950s based on their experimental research on the giant axon of the squid. This model includes conductances for sodium (Na), potassium (K), and leak currents, representing the opening degree of each ion channel. The opening and closing of the ion channels in the model are further described by gating variables, which are voltage- and time-dependent (m, h, n). The equations of the HH model are as follows:

Where is the membrane potential, is the input current, ​, ​, and are the maximum conductances for potassium, sodium, and leak currents, respectively. ​, ​, and ​ are the equilibrium potentials for potassium, sodium, and leak currents, respectively. , and are variables associated with the states of ion channel gating for potassium and sodium currents.
The dynamics can be described by the following differential equations:

The and functions represent the rates of channel opening and closing, which were experimentally determined using the patch-clamp technique.

Leaky Integrate-and-Fire Model (LIF)

The Leaky Integrate-and-Fire model (LIF) is a commonly used mathematical model in neuroscience that simplifies the action potential of neurons. This model focuses on describing the temporal changes in membrane potential [4-5] while neglecting the complex ionic dynamics within biological neurons.

Scientists have found that when a continuous current input is applied to a neuron [6-7], the membrane potential rises until it reaches a certain threshold, leading to the firing of an action potential, after which the membrane potential rapidly resets and the process repeats. Although the LIF model does not describe the specific dynamics of ion channels, its high computational efficiency has led to its widespread application in neural network modeling and theoretical neuroscience research. Its basic equation is as follows:

Where is the membrane potential; is the resting membrane potential; is the input current; is the membrane resistance; and is the membrane time constant, reflecting the rate at which the membrane potential responds to input current. is the membrane capacitance.

In this model, when the membrane potential reaches a specific threshold value ​, the neuron fires an action potential (spike). Subsequently, the membrane potential is reset to a lower value ​ to simulate the actual process of neuronal firing.

(2) Conduction of Electrical Signals in Neuronal Processes (Cable Theory)

In the late 19th to early 20th centuries, scientists began to recognize that electrical signals in neurons could propagate through elongated neural fibers such as axons and dendrites. However, as the distance increases, signals tend to attenuate. Researchers needed a theoretical tool to explain the propagation of electrical signals in neural fibers, particularly the voltage changes over long distances.

In 1907, physicist Wilhelm Hermann proposed a simple theoretical framework that likened nerve fibers to cables to describe the diffusion process of electrical signals. This theory was later further developed in the mid-20th century by Hodgkin, Huxley, and others, who confirmed the critical role of ion flow in signal propagation through experimental measurements of neurons and established mathematical models related to cable theory.

The core idea of cable theory is to treat nerve fibers as segments of a cable, introducing electrical parameters such as resistance and capacitance to simulate the propagation of electrical signals (typically action potentials) within nerve fibers. Nerve fibers, such as axons and dendrites, are viewed as one-dimensional cables, with electrical signals propagating along the length of the fiber; membrane electrical activity is described through resistance and capacitance, with current conduction influenced by internal resistance and membrane leakage resistance; the signal gradually attenuates as it propagates through the fiber.

▷ Fig.8: Schematic diagram of cable theory.

The cable equation is:

Where ​ represents the membrane capacitance per unit area, reflecting the role of the neuronal membrane as a capacitor; is the radius of the nerve fiber, which affects the propagation range of the electrical signal; ​ is the resistivity of the nerve fiber’s axial cytoplasm, describing the ease of current propagation along the fiber; and ​ is the ionic current density, representing the flow of ion currents through the membrane.

The temporal change, governed by the term CM and ∂ V(x,t) / ∂ t​, reflects how the membrane potential changes over time; the spatial spread, represented by the term (α/ 2 ρL)*(∂^2 V(x,t)/ ∂^2 x*2), describes the gradual spread and attenuation of the signal along the nerve fiber, which is related to the fiber’s resistance and geometry. The term​ iion indicates the ionic current through the membrane, which controls the generation and recovery of the action potential. The opening of ion channels is fundamental to signal propagation.

(3) Multi-Compartment Model

In earlier neuron modeling, such as the HH model and cable theory, neurons were simplified to a point-like “single compartment,” only considering temporal changes in membrane potential while neglecting the spatial distribution of various parts of the neuron. These models are suitable for describing the mechanisms of action potential generation but fail to fully explain signal propagation in the complex morphological structures of neurons (such as dendrites and axons).

As neuroscience deepened its understanding of the complexity of neuronal structures, scientists recognized that voltage changes in different parts of the neuron can vary significantly, especially in neurons with long dendrites. Signal propagation in dendrites and axons is influenced not only by the spatial diffusion of electrical signals but also by structural complexity, resulting in different responses. Thus, a more refined model was needed to describe the spatial propagation of electrical signals in neurons, leading to the development of the multi-compartment model.

The core idea of the multi-compartment model is to divide the neuron’s dendrites, axons, and cell body into multiple interconnected compartments, with each compartment described using equations similar to those of cable theory to model the changes in transmembrane potential  over time and space. By connecting multiple compartments, the model simulates the complex propagation pathways of electrical signals within neurons and reflects the voltage differences between different compartments. This approach allows for precise description of electrical signal propagation in the complex morphology of neurons, particularly the attenuation and amplification of electrical signals on dendrites.

Specifically, neurons are divided into multiple compartments, each representing a portion of the neuron (such as dendrites, axons, or a segment of the cell body). Each compartment is represented by a circuit model, with resistance and capacitance used to describe the electrical properties of the membrane. The transmembrane potential is determined by factors such as current injection, diffusion, and leakage. Adjacent compartments are connected by resistors, and electrical signals propagate between compartments through these connections. The transmembrane potential Vi follows a differential equation similar to cable theory in the i-th compartment:

Where Ci is the membrane capacitance of compartment, Imem,i(t) is the membrane current, and Raxial,i is the axial resistance between compartments. These coupled equations describe how the signal propagates and attenuates within different compartments of the neuron.

In the multi-compartment model, certain compartments (such as the cell body or initial segment) can generate action potentials, while others (like dendrites or axons) primarily facilitate the propagation and attenuation of electrical signals. Signals are transmitted through connections between different compartments, with input signals in the dendritic region ultimately integrated at the cell body to trigger action potentials, which then propagate along the axon.

Compared to single-compartment models, the multi-compartment model can better reflect the complexity of neuronal morphological structures, particularly in the propagation of electrical signals within structures like dendrites and axons. Due to the coupling differential equations involving multiple compartments, the multi-compartment model often requires numerical methods (such as the Euler method or Runge-Kutta method) for solution.

2.2.2 Why Conduct Complex Dynamic Simulations of Biological Neurons?

Research by Beniaguev et al. has shown that the complex dendritic structures and ion channels of different types of neurons in the brain enable a single neuron to possess extraordinary computational capabilities, comparable to those of a 5-8 layer deep learning network [8].

▷ Fig.9: A model of Layer 5 cortical pyramidal neurons with AMPA and NMDA synapses, accurately simulated using a Time Convolutional Network (TCN) with seven hidden layers, each containing 128 feature maps, and a historical duration of 153 milliseconds.

He et al. focused on the relationships between different internal dynamics and complexities of neuron models [9]. They proposed a method for converting external complexity into internal complexity, noting that models with richer internal dynamics exhibit certain computational advantages. Specifically, they theoretically demonstrated the equivalence of dynamic characteristics between the LIF model and the HH model, showing that an HH neuron can be dynamically equivalent to four time-varying parameter LIF neurons (tv-LIF) with specific connection structures.

▷ Fig.10: A method for converting from the tv-LIF model to the HH model.

Building on this, they experimentally validated the effectiveness and reliability of HH networks in handling complex tasks. They discovered that the computational efficiency of HH networks is significantly higher compared to simplified tv-LIF networks (s-LIF2HH). This finding demonstrates that converting external complexity into internal complexity can enhance the computational efficiency of deep learning models. It suggests that the internal complexity small model approach, inspired by the complex dynamics of biological neurons, holds promise for achieving more powerful and efficient AI systems.

▷ Fig.11: Computational resource analysis of the LIF model, HH model, and s-LIF2HH.

Moreover, due to structural and computational mechanism limitations, existing artificial neural networks differ greatly from real brains, making them unsuitable for directly understanding the mechanisms of real brain learning and perception tasks. Compared to artificial neural networks, neuron models with rich internal dynamics are closer to biological reality. They play a crucial role in understanding the learning processes of real brains and the mechanisms of human intelligence.

2.3 Challenges

Despite the impressive performance of the internal complexity small model approach, it faces a series of challenges. The electrophysiological activity of neurons is often described by complex nonlinear differential equations, making model analysis and solution quite challenging. Due to the nonlinear and discontinuous characteristics of neuron models, using traditional gradient descent methods for learning becomes complex and inefficient. Furthermore, increasing internal complexity, as seen in models like HH, reduces hardware parallelism and slows down information processing speed. This necessitates corresponding innovations and improvements in hardware.

To tackle these challenges, researchers have developed various improved learning algorithms. For example, approximate gradients are used to address discontinuous characteristics, while second-order optimization algorithms capture curvature information of the loss function more accurately to accelerate convergence. The introduction of distributed learning and parallel computing allows the training process of complex neuron networks to be conducted more efficiently on large-scale computational resources.

Additionally, bio-inspired learning mechanisms have garnered interest from some scholars. The learning processes of biological neurons differ significantly from current deep learning methods. For instance, biological neurons rely on synaptic plasticity for learning, which includes the strengthening and weakening of synaptic connections, known as long-term potentiation (LTP) and long-term depression (LTD). This mechanism is not only more efficient but also reduces the model’s dependence on continuous signal processing, thereby lowering the computational burden.

▷ MJ

3. Bridging the Gap Between Artificial Intelligence and Human Brain Intelligence

He et al. theoretically validated and simulated that smaller, internally complex networks can replicate the functions of larger, simpler networks. This approach not only maintains performance but also enhances computational efficiency, reducing memory usage by four times and doubling processing speed. This suggests that increasing internal complexity may be an effective way to improve AI performance and efficiency.

Zhu and Eshraghian commented on He et al.’s article, “Network Model with Internal Complexity Bridges Artificial Intelligence and Neuroscience” [5]. They noted, “The debate over internal and external complexity in AI remains unresolved, with both approaches likely to play a role in future advancements. By re-examining and deepening the connections between neuroscience and AI, we may discover new methods for constructing more efficient, powerful, and brain-like artificial intelligence systems.”

As we stand at the crossroads of AI development, the field faces a critical question: Can we achieve the next leap in AI capabilities by more precisely simulating the dynamics of biological neurons, or will we continue to advance with larger models and more powerful hardware? Zhu and Eshraghian suggest that the answer may lie in integrating both approaches, which will continuously optimize as our understanding of neuroscience deepens.

Although the introduction of biological neuron dynamics has enhanced AI capabilities to some extent, we are still far from achieving the technological level required to simulate human consciousness. First, the completeness of the theory remains insufficient. Our understanding of the nature of consciousness is lacking, and we have yet to develop a comprehensive theory capable of explaining and predicting conscious phenomena. Second, simulating consciousness may require high-performance computational frameworks that current hardware and algorithm efficiencies cannot yet support. Moreover, efficient training algorithms for brain models remain a challenge. The nonlinear behavior of complex neurons complicates model training, necessitating new optimization methods. Many complex brain functions, such as long-term memory retention, emotional processing, and creativity, still require in-depth exploration of their specific neural and molecular mechanisms. How to further simulate these behaviors and their molecular mechanisms in artificial neural networks remains an open question. Future research must make breakthroughs on these issues to truly advance the simulation of human consciousness and intelligence.

Interdisciplinary collaboration is crucial for simulating human consciousness and intelligence. Cooperative research across mathematics, neuroscience, cognitive science, philosophy, and computer science will deepen our understanding and simulation of human consciousness and intelligence. Only through collaboration among different disciplines can we form a more comprehensive theoretical framework and advance this highly challenging task.