Beginning some 35,000 years ago, hundreds of cave sanctuaries throughout southern France and Spain were lavishly adorned with beautiful and evocative paintings and engravings. Prehistoric artists carried out their work with remarkable stylistic continuity for over 20,000 years. Since this world of buried art was rediscovered and explored in the last hundred years, these paintings have been admired for their rich, expressive depictions of animals and geometrical patterns.

But what do these paintings mean, and why were they created? How were these caves used? At various times, scholars have interpreted cave paintings as art for art’s sake, hunting instructions, sympathetic magic, totemistic representations of clan identities, or symbolic vocabularies with complex systems of meaning.

In The Shamans of Prehistory: Trance and Magic in the Painted Caves, two prominent researchers argue that many European caves are linked to shamanic ritual practice and initiation. Renowned expert Jean Clottes, who served as principal researcher of the magnificent Chauvet cave of southern France, co-authored this book with South African cognitive archaeologist David Lewis-Williams, a specialist on the San culture of the Kalahari, which practices rock painting to this day.

At 120 pages, this book is essentially a long essay laying the basis for the authors’ shamanic hypothesis and attempting to ground it in biological terms. While I found their central thesis to be underdeveloped, the authors do an admirable job of surveying the available evidence, providing a valuable analysis of the known art. The book is lavishly illustrated, though the pictures are rarely captioned with date information.

In the book’s introduction, the authors present a brief account of shamanism as a religious paradigm. They focus on the role of shamans as expert practitioners who carry out supernatural feats by entering trance states through various means, including the use of visionary plants, isolation, fasting, chanting, and dancing. In these altered states, they travel into the heavens above, or into a world beneath the earth, where they encounter spirits and animal powers who assist them in their work.

The authors argue that the visionary states experienced by the shaman conform to a three-stage model that characterizes trance or altered consciousness, including those evoked by ritual practice and those caused by psychoactive substances such as LSD. The authors interpret the generality of their three-stage model as evidence for a shared biological process at work, one that is triggered in different ways but produces a similar experience.

Stage one of their model consists of the appearance of vivid, luminous, geometric patterns. In stage two, the fuzzy and ambiguous geometric images begin to take on meaningful shapes and symbols, as the subject “recognizes” them as outlines of known shapes (e.g. horses, lions, etc.). The transition between stages two and three is often marked by an experience of passage, such as moving through a tunnel or flying. Stage three involves frank hallucinations of otherworldly symbols and beings.

This model is used throughout the book as a framework to explain the universality of characteristic shamanic visions, such as the magical flight, which may be interpreted as the transition between stages two and three. Many cave paintings can be interpreted as reflecting one or more of the stages. This may indicate the caves were ritual or initiatory centers that either depict or help elicit shamanic visions brought about through various means.

Given the weight the authors place on their altered states model, I was disappointed to see it presented with very little evidence or justification. The model is supported by a single work, “The Signs of All Times. Entoptic Phenomena in Upper Paleolithic Art” (1988), an article co-authored by Lewis-Williams for the journal Current Anthropology.

In my opinion, the model is inadequate as an explanatory mechanism, and unable to do the heavy lifting Lewis-Williams and Clottes require of it. Its key terms are extremely vague—especially the central concept of “altered states”.

This term can refer to a vast array of states of awareness, including alertness, stupor, delirium, hallucination or bliss. Even if we restrict ourselves to visionary situations that involve both visual distortions and frank hallucinations, we still find a diverse set of experiences that is poorly characterized by this model. Comparing the effects of datura, Salvia divinorum, and psilocybin mushrooms, for example, we can immediately see that they don’t conform to a single model.

What we are left with, then, is a hypothetical biological process of unknown nature, one that somehow produces similar effects under specific but unknown conditions. And this is supposed to empirically ground their theory?

The authors suggest at several points that the fitness of their altered-states model to the evidence may indicate that hallucinogenic plants were ritually used. To evaluate that hypothesis, we need to examine which hallucinogens fit their three-stage model, and ask if they were available in Europe in prehistoric times.

I submit that the classical tryptamine and phenethylamine hallucinogens, such as psilocybin or DMT and mescaline, are the best fit for their altered-states model. Unfortunately, these are overwhelmingly found in the New World, and were probably unknown within Europe until sometime long after the caves had been painted.

What potentially hallucinogenic substances were most likely to be available in the late Stone Age in Europe? I suggest the following candidates: carbon dioxide, cannabis, opium, Amanita muscaria, Syrian rue, and solenaceous plants, including datura and belladonna.

Setting aside carbon dioxide for the moment, none of these are an obvious fit with the three-stage model. In particular, none of these substances would reliably produce the three stages of the model. We’re looking for something that causes an experience so familiar that it was the foundation of the art and religious practice in the region for 20,000 years.

At the right dose levels, carbon dioxide intoxication does fit well with the three-stage theory, as we learn from the extensive research of Dr. Ladislas Meduna. The authors do not mention carbon dioxide intoxication in this book, but Clottes speaks of it in his Cave Art (Phaidon, 2010), where he speculates that some cave chapels may have caused carbon dioxide intoxication due to poor ventilation and this could have played a role in the paintings.

The problem with this theory is that high levels of carbon dioxide rapidly cause unconsciousness and death; indeed, the gas is frequently used to euthanize animals. Hallucinations generally occur at the threshold of unconsciousness, and it’s hard to imagine how any shaman could fall insensibly into a visionary stupor in the depths of a cavern thick with carbon dioxide, and then live to tell the tale.

Psilocybin-containing mushrooms may have been known in Europe in prehistoric times, but the evidence for this is extremely tenuous. Surprisingly, the authors don’t consider the famous rock art bee-masked being that may be covered with mushrooms, found on the Tassili plateau of southern Algeria. But the link between that image and psychoactive mushrooms is speculative, and Algeria is a long way from Dordogne.

If there is a shred of evidence that Tabernanthe iboga was available in Europe in prehistoric times, I have not heard of it.

The altered states model is unpersuasive on aesthetic grounds as well. I have seen a great deal of visionary art that evokes states of ecstasy, and none of it bears any resemblance to the vast majority of what we find in cave art. Cave paintings usually depict easily recognizable animals in crisp, elegant outlines, either isolated or in small groups. Modern visionary art inspired by hallucinations, on the other hand, frequently emphasizes figure-ground ambiguity with crowded visual fields saturated with suggestive images.

It would also be remarkable to find a long-lived visionary bestiary so limited in its repertoire. We frequently find horses, aurochs, and mammoths, but almost never snakes, insects, or birds. What kind of visionary artist doesn’t paint snakes?

I don’t believe the theory works much better with endogenous altered states. Trance states evoked by meditation, chant, isolation, prayer, or dance are no less diverse than those evoked by psychoactive substances. I don’t see the three-stages model as a good description for my experience of any of them.

It’s entirely possible, or even likely, that psychoactive plants were part of the spiritual tool kit for Homo sapiens in the Paleolithic, but I don’t see clear evidence linking them to cave art.

I am also a firm proponent of the shamanism model for understanding Paleolithic cave art, but on comparative grounds, such as those advanced by Mircea Eliade. The structure of many cave sanctuaries strongly suggests an initiatory domain, easily recognizable from sacred spaces used by cultures today. The placement of key artwork in remote, difficult-to-access chambers implies a journey. The animal images are of an archaic character that fit extremely well with what we observe in contemporary shamanic cultures, such as among the Intuit, Tlingit, or Haida. And some of the composite “sorcerer” paintings are richly evocative of trance states or initiatory visions of a well-known type.

Clottes and Lewis-Williams, I believe, are trying to make a different kind of argument on behalf of the shamanic hypothesis, because comparative work is out of style in the academy. But they haven’t been persuasive here, and I believe the serious conceptual problems they encounter with their models are symptomatic of the weakness of this approach. It is extremely difficult to persuasively link high-scale cultural phenomena to low-scale neurobiological processes, and in this case the biological basis is merely hypothetical.