Boden is the cycle’s master diagnostician. Where other thinkers provide arguments about what the machines can and cannot do, she provides the framework within which the arguments can be precisely formulated. Her three-part taxonomy transforms “can AI be creative?” into three sharper questions with different answers. Her distinction between P-creativity and H-creativity explains why users oscillate between awe and disappointment with the same system: the same output is a genuine marvel from the user’s side (P-creative) and a sophisticated recombination from the engineer’s (traceable to training distribution). Her conceptual-space framework makes precise the claim that current systems excel at exploration within learned spaces and face structural limits in transforming those spaces.
Her most important contribution to the cycle may be her insistence that modeling is understanding: the attempt to build a thinking machine is one of the most powerful tools ever devised for understanding what thinking actually is. Every failure of a system at a specific task marks the boundary of what we have managed to formalize, and points at the vast territory of mind we have not. Where Pearl maps the limits of first-rung machine intelligence through the framework of causation, Boden maps the limits of machine creativity through the framework of transformation—and both arrive at the same conclusion from different directions: the machines have taken a great deal of human cognitive territory, and the territory they have not yet taken is exactly the territory we understand least.
Her temperament—marvel without mystery—is precisely what the cycle tries to model. She refused mystification (the insistence that creativity is sacred and ineffable, which protects wonder by forbidding understanding) and refused reductive triumphalism (the insistence that since the mind is computation and the machines now compute, there is nothing left to marvel at). She held that understanding deepens rather than diminishes wonder. The machines are astonishing. Understanding them does not make them less astonishing. It makes the astonishment honest. That is the orange pill applied to the creativity question.
Boden also provides the cycle with its most important historical corrective. Having lived through and helped write the entire history of AI, she knew that every previous paradigm had felt, at its peak, like the final answer. GOFAI once seemed to have captured all of intelligence; neural networks now seem to have. Her long-view warning is the cycle’s antidote to the temptation to declare the problem solved: the current systems have conquered pattern and association magnificently, and they remain on the near side of barriers that may require ideas we do not yet have.
Born in London in 1936, Boden took first-class honours in medical sciences at Cambridge in two years rather than three, then studied philosophy, then earned a PhD in social psychology at Harvard. The turn to cognitive science came when she read Miller, Galanter, and Pribram’s Plans and the Structure of Behavior (1960) and recognized that computational ideas could illuminate the whole of psychology. At the University of Sussex she built, with Aaron Sloman, one of the first schools in the world dedicated to cognitive and computing sciences as a unified discipline. Her 1977 book Artificial Intelligence and Natural Man was one of the first systematic analyses of what AI systems could and could not do and what their limitations revealed about the structure of human cognition.
Her major work on creativity—The Creative Mind: Myths and Mechanisms—appeared in 1990, before the current generation of generative systems existed, and it supplied a conceptual apparatus so well-designed that the systems it best illuminates are the ones built decades later. The two-volume Mind as Machine: A History of Cognitive Science (2006) is the definitive account of the field she helped found. Her final extended contribution, AI: Its Nature and Future (2016), applies her accumulated insight to the then-emerging generation of deep-learning systems with the same rigor and the same disciplined agnosticism about consciousness that characterized everything she wrote.
Three kinds of creativity. Combinational, exploratory, and transformational creativity are generated by different mental processes, answer to different descriptions, and are accomplished by machines to very different degrees. The taxonomy is the most useful single instrument for the question of machine creativity because it replaces a blanket yes or no with three answers that can be empirically investigated.
The conceptual space. A conceptual space is a structured domain of possibilities defined by a set of constraints. Exploratory creativity moves through the space; transformational creativity alters or abandons one of its defining constraints. Current generative systems are, almost literally, exploratory engines: they have learned a vast conceptual space from their training data and return positions within it. Their structural ceiling is the boundary of the space they learned—a machine that has learned the existing road system cannot easily build a road to a destination the old map could not represent.
P-creativity and H-creativity. Psychological creativity is novelty new to the mind that produces it; historical creativity is novelty new to the whole of human culture. The machines have P-creativity in abundance: they produce outputs routinely novel in the user’s personal experience. Whether they approach H-creativity in any domain is the contested frontier. The distinction explains the oscillation between awe and disappointment with AI output—the same system can satisfy both descriptions simultaneously.
Mind as virtual machine. The mind is to the brain as a spreadsheet is to silicon: a real, causally potent level of organization implemented in but not reducible to its physical substrate. This makes machine consciousness conceivable in principle without making it automatic. The hard question is whether the right virtual machine can run on silicon, and whether phenomenal consciousness depends on specifics of biological implementation that digital systems do not share. Boden kept this genuinely open for sixty years.
Marvel without mystery. The deepest temperamental gift Boden offers is the refusal of the false choice between mystification and disenchantment. Understanding a mind does not kill the wonder; it makes the wonder honest. This is the stance the cycle tries to model: rigorous astonishment, held without anxiety, in the face of systems whose capabilities are genuinely astonishing and whose inner nature remains genuinely uncertain.