DIKWP Scientific Predictions and Future Outlook

通用人工智能AGI测评DIKWP实验室

2025-11-01

Yucong Duan

International Standardization Committee of Networked DIKWPfor Artificial Intelligence Evaluation(DIKWP-SC)

World Artificial Consciousness CIC(WAC)

World Conference on Artificial Consciousness(WCAC)

(Email: duanyucong@hotmail.com)

Introduction

Based on the previous analysis of the DIKWP semantic mathematics framework and the consciousness "BUG" theory, as well as current progress in neuroscience, we propose the following scientific predictions and research directions. Each direction will be elaborated upon in detail, combining existing literature and neural mechanisms, and corresponding experimental or technical concepts will be proposed, aiming to provide testable hypotheses for consciousness and semantic research.

Measurable Neural Markers for "BUG"

If the "BUG theory" holds true, the generation of consciousness should correspond to observable neural events. For example, when the complexity of a cognitive task exceeds the brain's bottom-up processing capacity, the brain may need to invoke pathways such as the frontal-cingulate-thalamic pathway to produce sudden synchronous responses. Studies have shown that in visual conscious perception tasks, neural activity between the thalamus (especially the inner thalamic nuclei) and the prefrontal cortex is activated before consciousness emerges and produces synchronous coupling during conscious perception. Meanwhile, when subjects report seeing a stimulus, the corresponding P300 event-related potential is significantly enhanced. These results indicate that conscious perception is highly correlated with specific brain electrical activity characteristics. We thus predict: in a state of cognitive overload (e.g., when task difficulty exceeds the manageable limit), the frontal-cingulate-thalamic pathway will exhibit a transient synchronous burst (similar to an enhanced P300 waveform), corresponding to a sudden shift in the level of consciousness. Future research can verify whether the aforementioned synchronous neural markers appear during the decision-making or "anomaly processing" stages by designing high-load cognitive tasks combined with high-density EEG or cortical/deep electrode recordings.

Anatomical Separation of DIKWP Levels

From a brain structure perspective, different DIKWP levels may rely on different brain regions. Existing neuroscience evidence shows that the hippocampus is the core substrate for long-term memory and semantic memory. Hippocampal damage severely affects memory formation, and the formation and consolidation of semantic memory both depend on the hippocampus. Therefore, the neural basis of the "Knowledge" (K) layer most likely includes the hippocampus and related temporal lobe structures; if this layer is damaged, it will mainly manifest as a loss of functions such as memory and semantic retrieval. In contrast, the prefrontal cortex, especially areas like the ventromedial prefrontal cortex (vmPFC), is closely related to high-level wisdom and purpose functions. Studies have found that patients with vmPFC damage show significant deficits in moral judgment and decision-making behavior, indicating that this region is crucial for "Wisdom/Purpose" (W/P) levels, such as moral reasoning and goal setting. At the same time, the dorsolateral prefrontal cortex is mainly involved in executive functions, planning, and control. To verify the neural boundaries of each level, multi-region damage or stimulation experiments can be conducted: observe the differential effects on cognitive functions at each DIKWP level when different prefrontal, parietal, cingulate, and hippocampal regions are damaged, thereby examining the reproducible boundaries in brain structure between the intelligence/purpose layer and the knowledge layer.

Multi-scale Network Dynamics Models

Methods like dynamic causal modeling or machine learning can be used to construct multi-scale brain network models that reflect DIKWP transformations. Dynamic Causal Modeling (DCM) has long been used to characterize effective connectivity in distributed neural networks, estimating causal interaction parameters between brain regions by fitting experimental data. We can perform DCM analysis on the brain in resting and task states to explore whether different DIKWP stages correspond to different functional connectivity patterns; for example, whether the activation of specific pathways is observed during the "Knowledge → Wisdom" transformation. On the other hand, methods like the Hidden Markov Model (HMM) can decompose brain activity into discrete sequences of network states. Some studies have proposed using HMM models to represent resting-state and task-state brain activity as a series of different functional network states. Drawing on this idea, a brain network Markov chain can be constructed, mapping different semantic processing stages to transition actions between network states. Furthermore, these models can be validated through perturbation experiments: for example, applying transcranial magnetic stimulation (TMS) or electrical stimulation to specific nodes and observing whether the semantic processing pathway changes as expected, to test whether the model's predictions of state transitions hold true. This approach of multi-scale modeling, combining Markov chains and DCM, is expected to characterize the information flow patterns between different DIKWP levels from a network dynamics perspective.

Comparative Experiments between Artificial Intelligence and Human Brain Semantics

Based on relative consciousness theory and semantic mathematics, we can design experiments that compare the processing of AI systems and the human brain on the same task. The specific method is: have an AI (such as a large language model) and a human answer the same question simultaneously, while recording the human brain's neural activity (such as fMRI or EEG data) during the answering process. Then, analyze the internal states of the AI model at various hidden layers and correlate them with the activity of different cognitive levels in the human brain. Some research has evaluated the correlation between various advanced Transformer models and human brain activity, finding that the RoBERTa model has the highest consistency with human brain activity when simulating text semantic processing, surpassing models like BERT. This suggests that the internal representation levels of artificial models may be comparable to different semantic processing stages in the human brain. If the DIKWP model is correct, we must not only verify the consistency of the final answers but also check whether intermediate steps, such as information extraction and knowledge application, have similar hierarchical alignment in AI and the human brain. However, cautious interpretation is also needed: studies have pointed out that neural networks without biological constraints often require the artificial addition of specific structures to reproduce neural patterns found in the brain. Therefore, these differences must be considered when using AI models as analogies for human brain functions. Through such comparative experiments, we can reveal the similarities and differences between human brains and artificial systems in multi-level semantic processing and provide new perspectives for validating semantic mathematics models.

Research on Collective Consciousness (Noosphere)

Inspired by the concept of collective consciousness (Noosphere), the neural basis of group collaborative cognition can be studied using EEG network hyperscanning technology. Specifically, in team cooperation or online interaction environments, the brain activities of multiple individuals (e.g., parallel EEG) can be recorded simultaneously to analyze the synchronization patterns between individual brain networks. Studies have found that in joint social or collaborative tasks, inter-brain synchrony occurs in the brain activity among team members, and the degree of this synchrony can predict the team's overall performance. For example, in a team problem-solving experiment, the EEG synchrony among team members was significantly positively correlated with the team's collective performance, whereas traditional self-attribution measures failed to effectively predict team performance. Based on this, further exploration can be conducted on real-time brainwave interactions in online gaming teams or social networks: when multiple people collaborate to solve a problem, will new synchronous hierarchical structures emerge, similar to a group-level "collective consciousness" phenomenon? If cross-individual brain network coupling is found during group collaboration, it indicates that a neural mechanism similar to a collective mind may exist. Such research not only provides new tools for understanding the neural dynamics in social collaboration but also validates the existence of collective cognitive states.

Bio-Artificial Fusion Systems

Applying DIKWP semantic mathematics to the field of Brain-Computer Interfaces (BCI) allows for the exploration of constructing high-level semantic channels between the brain and artificial systems. We envision using BCI to read the activity of specific networks in the individual's brain in real-time: for example, the Default Mode Network (DMN) or the prefrontal cortex, to attempt to decode the current "Purpose/Goal P". It has been suggested that the DMN constructs holistic situational representations by coordinating parallel activities across multiple brain regions; thus, the activity patterns of the DMN may map to the individual's current semantic context or purpose. Transmitting the decoded purpose information to an AI system can guide its subsequent goal-setting or decision-making; conversely, the AI system can also feed back the generated knowledge and wisdom layer information to the brain: by stimulating or providing feedback to the corresponding neural circuits (such as the prefrontal and parietal lobes), the brain's acceptance and integration of new information can be enhanced, achieving a more natural human-machine semantic fusion. However, it must be pointed out that decoding implicit cognitive information such as attention, purpose, and decisions from brain signals remains challenging. Therefore, establishing such an advanced semantic BCI channel requires combining advanced machine learning algorithms and extensive training data to continuously improve decoding accuracy and timeliness, ultimately achieving close collaboration between humans and AI at the level of thought.

Conclusion and Outlook

In summary, based on the analysis of DIKWP semantic mathematics and the consciousness "BUG" theory, we have proposed several cutting-edge research directions, from neural marker detection, anatomical layer verification, and network dynamics modeling, to AI comparative experiments, collective cognition research, and brain-computer interface applications. These predictions cover multiple aspects of the origin of consciousness, semantic processing, and human-machine interaction. By combining modern neuroscience technology and artificial intelligence methods, we hope to build new connections between experimental evidence and theoretical frameworks, thereby gaining a deeper understanding of the essence of consciousness and semantics. These explorations will not only enrich human understanding of brain function and consciousness but also point the way for the future development of active intelligence and brain-machine fusion systems.

References

Key Brain Gateway to Conscious Perception Identified - Neuroscience News, https://neurosciencenews.com/thalamus-conscious-perception-28545/

Does the P300 reflect conscious perception or its consequences? - PubMed, https://pubmed.ncbi.nlm.nih.gov/25907442/

“双面”海马体——海马体对空间和抽象知识的表征 (Multi-faceted Hippocampus - Representation of Spatial and Abstract Knowledge by the Hippocampus) - 脑医汇 (Brainmed.com), https://www.brainmed.com/info/detail?id=23365

Damage to the ventromedial prefrontal cortex is associated with impairments in both spontaneous and deliberative moral judgments - PubMed, https://pubmed.ncbi.nlm.nih.gov/29382558/

Dynamic Causal Modeling on the Identification of Interacting Networks in the Brain: A Systematic Review - PubMed, https://pubmed.ncbi.nlm.nih.gov/34714747/

Discovering dynamic brain networks from big data in rest and task - UK Biobank, https://www.ukbiobank.ac.uk/publications/discovering-dynamic-brain-networks-from-big-data-in-rest-and-task/

[2501.06278] Aligning Brain Activity with Advanced Transformer Models: Exploring the Role of Punctuation in Semantic Processing, https://arxiv.org/abs/2501.06278

Study urges caution when comparing neural networks to the brain | Brain and Cognitive Sciences, https://bcs.mit.edu/news/study-urges-caution-when-comparing-neural-networks-brain

Inter-brain synchrony in teams predicts collective performance - PubMed, https://pubmed.ncbi.nlm.nih.gov/32991728/

How Does the “Default Mode” Network Contribute to Semantic Cognition? - PMC , https://pmc.ncbi.nlm.nih.gov/articles/PMC11135161/

Frontiers | Beyond the brain-computer interface: Decoding brain activity as a tool to understand neuronal mechanisms subtending cognition and behavior, https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2022.811736/full