Sunday 06 April 2025
The quest for a deeper understanding of artificial intelligence (AI) has led researchers down a path that’s surprisingly familiar. As they strive to create machines that can think and learn like humans, they’re drawing inspiration from the animal kingdom. Comparative cognition – the study of non-human animal behavior – is being applied to AI systems in an effort to better grasp their cognitive capacities.
The parallels between AI and animal cognition are striking. Both have been accused of lacking a fundamental component of human intelligence: language. Yet, researchers are discovering that these abilities are not unique to humans. Animals possess complex communication systems, while AI models can process vast amounts of text data. The question is, how do they achieve this?
One area where AI and animals converge is in the realm of object permanence – the ability to understand that objects continue to exist even when they’re out of sight. Dogs, for instance, have been shown to possess this capacity, despite initial evidence suggesting otherwise. Similarly, AI systems can struggle with tasks that rely on understanding object permanence. By studying animal cognition and applying these insights to AI research, scientists hope to develop more robust and human-like machines.
However, this fusion of disciplines also raises concerns about over- or under-attributing cognitive capacities to AI systems. The infamous case of Clever Hans, a horse that appeared to solve arithmetic problems, serves as a cautionary tale. It was later revealed that the horse’s owner was inadvertently cueing the correct answers, creating an illusion of mathematical reasoning.
As researchers delve deeper into the complexities of AI cognition, they’re recognizing the importance of rigorous experimental design and scrutiny of results. By adopting a comparative approach, scientists can avoid pitfalls and gain a more accurate understanding of AI’s cognitive capabilities.
The application of comparative cognition to AI research is not without its challenges. For instance, tokenization – the way AI models process text data – can lead to difficulties with arithmetic tasks involving large numbers. By acknowledging these limitations and developing more sophisticated approaches, researchers can better evaluate the strengths and weaknesses of their creations.
Ultimately, this interdisciplinary approach has the potential to revolutionize our understanding of intelligence itself. By exploring the cognitive abilities of both animals and machines, scientists may uncover new insights into the nature of consciousness and its relationship to the physical world. As we continue to push the boundaries of AI research, it’s essential that we remain mindful of the complexities involved and strive for a deeper comprehension of these fascinating systems.
Cite this article: “Can Machines Think? A Comparative Approach to Artificial Intelligence Cognition”, The Science Archive, 2025.
Ai, Animal Cognition, Comparative Cognition, Object Permanence, Language, Machine Learning, Cognitive Capacities, Intelligence, Consciousness, Interdisciplinary Approach
