Impact of technical reasoning and theory of mind on cumulative technological culture: insights from a model of micro-societies

Cumulative technological culture (CTC), the continuous improvement of tools over generations, is a defining feature of human success. While CTC has been studied extensively using mathematical and computational models focusing on factors like demographics, social learning strategies, and population structure, the specific cognitive skills driving it remain unclear. Two prominent hypotheses exist: the theory of mind (ToM) hypothesis, emphasizing the role of social-cognitive skills in faithful imitation; and the technical reasoning (TR) hypothesis, highlighting the importance of understanding the physical world in tool use and improvement. This paper aims to address this gap by developing a computational model integrating both ToM and TR within a micro-society framework. This micro-society paradigm allows for the study of knowledge transmission and technological improvement across generations under different social learning conditions: reverse engineering (RE), observation (OBS), and communication (COM). These conditions vary in the amount of information provided to the learner, with COM providing the most information. The model focuses on a single technology with multiple traits, each having a quality and a limit, allowing for exploration of both refinement and innovation in technology development. The model allows for precise manipulation of social-learning conditions and extends beyond the limitations of real-world experiments by simulating hundreds of generations.

Existing literature on CTC primarily uses mathematical and computational modeling to explore macroscale factors, neglecting the role of specific cognitive skills. The ToM hypothesis posits that faithful transmission of technical knowledge, facilitated by ToM, is crucial for CTC. However, recent studies using micro-society paradigms challenge this, showing cumulative performance even in RE conditions where only the final product is observed. These studies suggest the technical-reasoning hypothesis, proposing that TR, the ability to reason about physical object properties, is key to CTC. TR enables both analogical and causal reasoning about tool-object relationships. The micro-society experiments that form the foundation for the model examine the transmission of technological skills in three social learning forms (RE, OBS, and COM). A key challenge is the lack of standardized definition for OBS. The model incorporates a synthesis of nine experimental studies using the micro-society paradigm to ensure consistency with empirical data. These studies reveal that TR, particularly in the receiver, is a major predictor of cumulative performance.

The model simulates a micro-society with a single technology composed of multiple traits, each having a quality and a maximum achievable quality (limit). Individuals in each generation learn through RE, OBS, or COM, influenced by their own TR and ToM and those of the sender. TR is drawn from a Gaussian distribution based on the environment’s technological quality, representing interaction with the physical world. ToM is drawn from a Gaussian distribution, ranging from 0 (no ToM) to 1 (perfect ToM). Learning modifies an individual's knowledge of each trait. RE only considers the technology's current state, OBS combines RE with observed changes, and COM incorporates RE, observed changes, and sender's ToM, weighted by the receiver's TR. Individuals can improve or deteriorate the technology, with the probability of improvement related to their TR. Innovation is introduced as a chance event, with success depending on either high TR or near-optimal technology quality, adding new traits and increasing the limits of existing ones. The model is tested under four scenarios: (1) refinement only; (2) refinement and innovation; (3) focusing on a single transmission chain to study detailed behavior; and (4) comparing models with and without ToM. Simulations run 1000 generations with varying parameters, and technology quality is averaged across multiple runs. Random seeds are used to control randomness in the analysis of single transmission chains. Parameters such as the technology limit, number of traits, innovation strength, improvement/deterioration factors, and the probability of innovation, are varied and observed.

The model's results align with micro-society literature, where COM consistently outperforms RE and OBS in technology quality. In the refinement-only scenario, COM reaches the optimal technology level faster and more reliably than RE and OBS. The differences in plateau levels are due to COM's superior ability to maintain high-quality technology over time. With innovation, all forms show exponential technology growth, but COM significantly outperforms others. RE and OBS show a pronounced decrease in technology quality when the deterioration factor is greater than the improvement factor. When increasing improvement factor, RE and OBS improve considerably, while COM has already reached its limit and does not improve. This suggests a snowball effect where improved technology leads to better understanding which in turn further improves technology. A single simulation run shows COM alternating between long periods of stasis (reaching the technology limit) and rapid growth after innovations, a pattern observed in human cultural evolution. RE and OBS exhibit similar behavior but with less frequent periods of innovation. Finally, the model reveals that ToM accelerates the initial growth of technology quality but is not essential for CTC, mainly impacting the early generations and not affecting long-term CTC. Micro-societies with COM showed resilience to cultural loss even in unfavorable conditions, possibly because TR facilitates understanding and correcting damage.

The model supports the importance of TR in CTC. TR facilitates both information transmission and technological modification. While ToM is not strictly essential, it acts as a catalyst, particularly in early stages. The results challenge the notion that faithful transmission requiring ToM is a prerequisite for CTC. The model aligns with existing micro-society studies and provides a detailed account of the mechanisms of cultural transmission and its dependence on cognitive skills. The model also demonstrates the impact of innovation on CTC, showing exponential growth under conditions including innovation, mirroring trends observed in human history. The observed alternation of rapid growth and stasis within single societies points to a mechanism for the overall exponential growth of culture that appears in populations of these societies.

This study demonstrates the crucial role of TR in driving CTC, highlighting its importance in both information transmission and technological innovation. ToM, while not essential, accelerates early growth. Future research could investigate innovation in more depth and explore the interaction of multiple technologies within a more complex environment. Incorporating additional cognitive abilities (such as altruism, creativity, and memory) would further refine the model's understanding of CTC.

The model simplifies reality by focusing on a single technology and neglecting other cognitive skills. The definition of OBS is challenging. The independence of technological traits might be considered an oversimplification, as interactions between traits are undoubtedly important in real-world scenarios. These simplifications, however, allow for a clear focus on the core mechanisms.