Friday, April 05, 2024

Merlin Donald - evolution, culture and ezternal sketchpad

Merlin Donald is a Canadian cognitive psychologist at Case West Reserve University, with a specific reputation in cognitive evolution, the evolution of the human mind. His particular focus is on the role of culture and technology in shaping human cognitive development, best known for his theory of cognitive evolution, which proposes that human cognition has evolved through several key stages, including episodic, mimetic, mythic, and theoretic culture.

He has extensively studied and written about consciousness, memory, and the neural basis of human cognitive abilities. He is also noted for examining the relationship between the mind and culture and how external symbolic systems such as language and art have influenced cognitive development.

Three Stages in Evolution and Culture

His book Origins of the Modern Mind: Three Stages in the Evolution of Culture and Cognition (1991), is a seminal work in the field of cognitive psychology and evolutionary psychology. Fior Merlin, human cognition has evolved through major stages:

1. Episodic Culture

Characterised by a reliance on episodic memory, where information is remembered as personal experiences without much abstract or symbolic representation.

This earliest stage of cognitive evolution is has a style of thinking and memory similar to that of non-human primates. It is called 'episodic' because it is primarily based on episodic memory, which is the recall of specific events and experiences, without the capacity for abstract reflection on these events. Numerous animal species seem to possess episodic memory, or the ability to recall specific past events along with the sights, sounds, and smells associated with them. Donald has labelled the early culture of hominids as episodic, paralleling that of chimpanzees. These early hominid communities, comparable in size to chimpanzee groups, used tools, imitated conspecifics with tools, and shared similar vocal apparatuses, with chimps producing up to 35 distinct vocalisations.

In this stage, humans would perceive and react to the world directly and immediately, much like other animals do. This stage is evidenced by early human fossils and the behaviour of modern primates. The absence of symbolic artifacts and tools from early human archaeological sites suggests a lack of symbolic thought, which is a key component of later cognitive stages.

2. Mimetic Culture

Humans developed the capability for mime or gesture, which enabled them to communicate more effectively and start building social cultures. This stage laid the foundation for the development of language.

This stage represents a significant leap in human cognition. It involves the development of mime or gesture, which enabled early humans to communicate through imitation and performance. This allowed for the transmission of skills and knowledge without relying solely on genetic inheritance. Mimetic culture laid the groundwork for more complex forms of communication, such as language, and represents the first form of cultural learning and transmission in human evolution.

'Mimetic' here doesn't refer to simple mimicry or imitation with adaptation. It signifies the capacity to use symbols for representation, particularly in communication. This opened the door to enhanced communication through vocalisations and gestures, more intricate social structures, and refined tool-making and hunting practices – hallmarks of Homo erectus. This species not only exhibited a significantly increased encephalisation quotient (EQ) but also crafted complex tools, harnessed fire, and traveled extensively. The emergence of more sophisticated tools and evidence of cooperative hunting and gathering in archaeological records suggest enhanced communication and social organisation. Cave paintings and other forms of prehistoric art provide indirect evidence of the ability to represent experiences through mime and gestures.

3. Mythic Culture

The advent of speech and language, allowing for the creation and transmission of complex narratives, myths, and cultural knowledge.

This phase sees the assimilation of segmented mimetic knowledge into unified and comprehensive narratives shared within communities. Such integration of knowledge facilitated its reorganisation and problem-solving. Homo sapiens, associated with mythic culture, innovated by creating new tools like needles, garments, and cave art. Homo sapiens also presented marked changes in skull and jaw structure, indicative of a developed vocal apparatus conducive to speech – which is speculated to have arisen even earlier.

The advent of spoken language marks this stage. The advent of spoken language marks this stage. With language, humans could create and transmit complex stories, myths, and collective knowledge. This allowed for the formation of larger social groups and more complex societal structures. Mythic culture is defined by the ability to narrate and construct shared myths, which served as the foundation for early forms of education, religion, and social cohesion. With language, humans could create and transmit complex stories, myths, and collective knowledge. This allowed for the formation of larger social groups and more complex societal structures. Mythic culture is defined by the ability to narrate and construct shared myths, which served as the foundation for early forms of education, religion, and social cohesion.

The evidence for this stage is found in ancient texts, oral traditions, and the sudden explosion of cultural artifacts in the archaeological record. The development of different languages and the complexity of ancient myths and stories reflect the cognitive abilities to think abstractly and symbolically.

Theoretic Culture

We are now in a ‘Theoretic Culture’, marked by the development of writing, formal systems of knowledge, and scientific thinking. The last cultural shift is marked by the documentation of symbolic forms on clay, wax tablets, papyrus, parchment, paper, and later on digital media. Human cognitive capacity, despite advanced brain function and is restricted by short-term memory, limiting the simultaneous processing and retention of information. This constraint makes recalling extensive knowledge to address intricate problems challenging. The invention of writing, particularly through the use of an alphabet, revolutionised this aspect.

His theory emphasizes the role of culture and external symbolic systems in shaping human cognition and posits that each stage of cognitive evolution builds upon and integrates with previous stages, leading to the complex and diverse forms of cognition we observe in humans today.

These stages highlight the gradual evolution of human cognition, emphasising the role of culture and symbolic thought in shaping our minds. Donald's theory suggests that each stage builds upon and integrates with the previous stages, leading to the complex forms of cognition we observe in humans today. This model has profound implications for understanding not only our past but also the nature of contemporary human thought and culture.

Merlin Donald, a cognitive psychologist, has made significant contributions to understanding human cognition and memory, particularly through his "Sketchpad Theory." This theory, more accurately referred to as his theories on cognitive evolution and external memory systems, emphasises the role of external symbolic systems in human memory and cognition.

External sketchpad

He argues that human memory and cognition have co-evolved with the use of external symbolic systems, such as art, writing, and other media. These external systems function as ‘memory storage’, an external sketchpad that enhances and extends our cognitive capacities.

This Theoretic Culture stage sees the development of external symbolic systems, such as writing, which allowed for more complex and abstract forms of knowledge representation. This is where his idea of an ‘external memory field’ or ‘external sketchpad’ arises, suggesting that these external systems are not just record-keeping tools but integral to the thought process itself.

He emphasises that human cognition is deeply intertwined with our interaction with external symbols. These systems do not just store information; they shape the way we think, reason, and remember. This is in line with the idea of the ‘extended mind’ put forward by Clark and Chalmers.


He drew on the ideas of Jean Piaget, Lev Vygotsky, and Jerome Bruner in the field of cognitive psychology. Both he and Geary have published similar books with similar titles but they come at the problem from different perspectives. Donald’s book was published before Geary’s but oddly, he fails to cite Donald’s work. Yet Donald’s work has continued to be influential for those who see brain science as being hugely influenced by the simple fact that it is an evolved organ.


The archaeological and history of our species have both moved on human cognition likely involves more overlap and interaction between these stages than the theory suggests. For example, oral traditions and mimetic skills continue to coexist alongside literate and digital cultures, suggesting a more integrated and less linear progression. His theories may underestimate the complexity of non-human cognition and some critics argue that Donald’s theory simplifies the complexity of cognitive processes. Human cognition involves a myriad of processes that are deeply interconnected, and reducing them to a sequence of evolutionary stages might not capture the full scope of these complexities.

Wednesday, April 03, 2024

Turchin's terrifying predictions

Peter Turchin rose to fame when his 2010 Nature article predicted the unrest in the US in the early 2000s. His book ‘End Times’ has created waves in the commentariat, as we head towards tricky elections in the US, UK and Europe this year. 

He brings science to history, not just the usual narrative approach, using huge data sets that flag up key indicators for political instability:

1. Decline in real wages - Tick
2. Growing gap between rich and poor - Tick
3. Overproduction of graduates - Tick
4. Declining public trust - Tick
5. Exploding public debt - Tick

As the ‘wealth pump’ pushes money to the top of the pyramid, the elite, wealth is pushed from the poor to the rich, accompanied by the disappointment of even middle class aspirants (graduates), who have also become part of the precariat.

Graduate class

The poor have suffered badly here (more later) but the credentialed class are also being left behind, as the pyramid stretches upwards and wealth accumulates at higher and higher levels.

For graduates it has become a game of ‘musical chairs’ where you pay a huge sum to buy a ticket to play the game (University costs), but the number of chairs (graduate level jobs) remain the same. As the number of players increases year on year, massively in just a few decades, supply way exceeds demand. So graduates have to up their game and pay for another even more expensive ticket to get a Masters. Even worse, those who go on to do PhDs find there are no academic jobs available, as again supply has exceeded demand for many years. Graduates in the social sciences and humanities are particularly vulnerable but Turchin’s point is that, for a rapidly increasing graduate population there is a precarious future and lots of debt. This may be exacerbated by AI, as it eats into cognitive work, so has that group as its sweet spot.

This frustrated aspirant class, for Turchin, is dangerous. Always isolated from working class people, they have little in common with the non-graduate class or ideas like collective bargaining and trade unions. They have the time and support from their propertied parents to become activists and protestors and often pick up causes on campus around cancel culture, climate change, transgender issues and social justice. Poverty is not the problem, recognition of identity is.

His hypothesis predicts a battle royale among the aspirant graduate class for declining opportunities and rewards. We have seen falls in enrolment in the US of 13 consecutive years, a flight from the humanities and negative attitudes towards Higher Education in the face of rapidly increasing costs. In an interesting addition. 

Turchin points out that cheating has also risen, with the sharp elbows of the middle class become ever more desperate to get their children into the right ‘brand’ colleges. Non-attendance at lectures and schools has also risen from the disengaged, disaffected and disconnected. The social contract seems to be breaking down, as the precariat realise that education is no longer a route to social mobility. It may even harm your chances, as you do not earn for many years and accumulate debt

Working class (Deplorables)

More seriously, there is a break in how rewards are distributed with the poorer getting poorer, that’s the 64% in the US without a college degree. His data on this is exemplary, along with his analysis of how the graduate class has turned their back on these people  - the famous ‘deplorables’. So the split economically and culturally, between the graduate and working class gets more and more extreme. What’s more, the social mobility of the working class has got worse as the challenge of paying for their children’s college education has moved further away from them. The wages of the educated with graduate jobs (the professions) grew, while the wages of the uneducated shrank. 

They were the victims of austerity after the banks collapsed, socialism for the rich – capitalism for the poor. Globalisation meant outsourcing to cheaper countries or importing cheaper labour. Collective bargaining was undermined and corporates restructured to save costs by shipping out, moving location or automation. Of course, the winners claim it was all down to their effort. Bullshit, it was. It was engineered by that same graduate class from their graduate business schools and MBAs. The professional classes, in positions of power across the entire system in politics, private schools, higher education, media and the arts redesigned the system around themselves and took it all.

And that, my friends, is how we got Trump… and Brexit… and Milei… and Meloni… and Wilders… and a swing to the far right across the whole of Europe from Spain to Sweden and Finland. 

I am not wholly convinced by the pure historicism, although he avoids simplistic cyclical theories, such as Schumpeter or political dialectics of Marxism. But what impresses is Turchin’s marshalling of data and wins on prediction and there is a LOT more in this book, especially his historical examples.

What is worrying is how easily we all march lock-step into the future, even when the signs of discontent are ringing in our ears. We are like those cartoon figures who run off the edge of the cliff and hang smiling in mid-air, before the fall. We don't adjust or adapt, we simply behave according to the groupthink of the socio-economic group we find ourselves in. The trick is to sit back, look, listen and read people like Turchin. You don't have to agree with him but it is voices like his that at least provide substance to predictions, not about 10nyears from now but next year!

The jury may be out but if he continues to predict crimes, the jury may turn out to be superfluous.

Tuesday, April 02, 2024

New psychology – Connectionists

The new psychology is the dangerous idea that traditional philosophy, psychology, psychotherapy and neuroscience has got it largel wrong. Connectionists see their theories grounded in the idea of the brain as a biological computer, operating computational processes, not the same as but showing similarities and parallels to nonbiological computing systems. They draw their primary inspiration from the work done from Hebbes onwards, in seeing the brain as a neural network. Both AI and the brain have as their basis huge, collective, connected networks.

They claim that ideas of the flat mind, with no hidden depths and computational models of the mind, fractures the self and the stories it tells itself, in favour of a more dynamic predictive improvising process. It is early days but already there are several serious thinkers who have emerged, building on existing thought, inspired by recent successes in artificial intelligence.

It is not Behaviourism but has something in common with that approach. Behaviourism, school of thought is often seen as being stupidly narrow and blunt in focussing upon observable behaviour only, ignoring the richness of internal reflection and experience. This is a caricature and they were far from stupid. There saw introspection as suspect and self-reported ideas about cognition unreliable. The new psychology is similarly suspicious, seeing such introspection as fatally flawed, especially when it involves language, as the mind creates stories and narratives about things that may not exist. Indeed, that is what we have evolved to do.

But they differ from the Behaviourists in having a much more substantial idea of how the brain works based on evidence from philosophy, neuroscience, mathematics and artificial intelligence. This is not a world of stimuli and responses but a complex interaction between the brain and world, which is predictive, inventive and improvisational. It is, like Behaviourism, critical of the explosion of theoretical constructs around 

This is an important movement, as it takes evidence from philosophy, psychology, neuroscience, then the computation models that took root in the 1950s leading to recent advances in AI to do nothing less than redefine what it is to be human. It also brings together what is happening in our brains and in our use of language. It is nothing less than a refresh.

This is more of a fledgling school of thought, certainly accelerated by the success of Generative AI, but by no means fully fledged in terms of evidence. Many claim it underestimates the complexity and depth of human thought, particularly in areas like abstract reasoning, long-term planning, and deep emotional experiences. Neither is it clear that it can account for the rich internal lives that individuals report and the extensive research supporting the existence of complex mental structures and processes.

The movement focuses more on System 1 than System 2 but confirm what Kahneman and others say about this, that System 1 is dominant. Note that what is being achieved in AI is not al LLMs and generative AI. There are also advanced in AI along planning and reason, using the ‘energy’ model put forward by Karl Friston to optimise inference, also the work of Josh Tenenbaum.

Nevertheless, as a cross-disciplinary effort, it is gathering momentum with its explanatory breadth. We can learn a lot about what the brain is, how the brain learns and how we perceive, operate in the world and use language by considering what is clearly assumed to be a computational theory of the mind. In creating something inspired by our own minds, we have created entities that seem, not only to do what we do but in cases surpass us. It may well turn out to be a theory that sees us as cognitively capped, compared to systems with a different substrate that escape those limitations. It is vital that we study and consider these issues as artificial intelligence progresses, as it surely will.

AI inspired psychology

This is a new frontier in psychology. The trajectory is towards a computational theory of the mind; a neural networked, intentioned, seeking, Bayesian, predictive brain. The group denies deeper cognitive entities such as the subject, self or soul as well as cognitive psychotherapeutic structures in favour of a flatter, shallower, more dynamic, adaptive and improvisational forms of cognition. It takes its inspiration from Enlightenment empirical philosophy, such as Hume who sees the self as a fiction throwing psychology back to the flow of experience by a perceiving brain, a theory refined by Parfit and Strawson. 

From the early 20th century, Hebb took inspiration from firing and wiring neurons to give us a theory of learning in the brain. This was given mathematical and logical form as neural nets by McCulloch and Pitts, an idea was turned into a working, learning perceptron by Rosenblatt and the race was on to find a way to make networks learn and perform in an intelligent fashion. Rumelhart and Hinton were to use backpropagation and layered neural networks along with LeCunn and Hassabis to push into Deep Learning. Large Language Models and their Multimodal offspring have produced astounding results that suggest we are track to at least find explanations that at last explain how the brain works.

During the 20th century there were also philosophers, neuroscientists and theorists that looked at language in detail, including Vygotsky and Wittgenstein, some with a focus on dialogue, such as Bakhtin and Pask. This is another area of inquiry that has been opened up with success in AI. Large Language and other multimodal models sprung into life, trained by enormous data sets, to predict a flow of language that seemed remarkably human, certainly useful. Then multimodal functionality appeared, where images, video and 3D worlds were used as both inputs and outputs and similarly successful generative capabilities were realised. This has confirmed the power of the connectionist approach. A number of connectionist theorists, now see cognition in a different way, more networked, diffuse, intentional, Bayesian, flat, active and predictive.

More recent computational theories of the mind have been influenced by the success of predictive, neural networks in artificial intelligence, with Dennett, Chater, Friston, Tenenbaum and Clark, among many others. They see the brain, not as a passive, receiving organ but as an active, predictive process.

The dialectic continues between the brain seen as a computational model and artificial intelligence which is definitely a computational model. We are seeing an explosion of progress and ideas on both fronts.

Flipped psychology

In a sense, the connectionist movement ‘flips’ psychology, seeing the brain as an evolved and far more active, dynamic, mercurial and improvising entity. Framing their work in the computational theory of mind, they call upon what we know about optimisation in AI to explain perception, thinking, language and consciousness.

Full conscious mind as receptive subject

Flatter computational mind, intentional and predictive 


Consciousness is an output, the end result of inferencing, like the output from most digital technologies, from the display of a calculator to the generative output of an LLM. The computational processes are black-boxed, unconscious and not part of consciousness. What we see and understand when we read or hear speech speak are the end results of a nexus of computational processes taking place in our 100 billion neurons. We could never be conscious of such processes, no more than being conscious of the vast network of data servers, the internet, electricity production and transmission and processes taking place in the circuitry of television, is perceived when using generative AI. We don’t see this, as there is nothing to ‘see’. What we can do is understand the underlying mathematics of the process.

Clark on AI, extended mind, neurodiversity and psychosis

Andy Clark is a British philosopher and cognitive scientist, Professor of Logic and Metaphysics at the University of Edinburgh. He has taught at the University of Sussex, the University of Washington in Seattle and Indiana University in Bloomington.

Particularly known for his research on the theory of the extended mind, which he developed with David Chalmers, he takes an interdisciplinary approach to the philosophy of mind, cognitive science and artificial intelligence, with an interest in predictive perception.

Extended mind 

Andy Clark and David Chalmers paper The Extended Mind (1998), explored the idea that the mind extends beyond the human brain to include our tools and environment. They hold that objects in the environment can become part of the mind, extending cognitive processes beyond our physical bodies. This concept has had a significant impact on philosophy, cognitive science, and discussions about the nature of consciousness. It is here they introduce the 'parity principle', where if an external object performs a function that, if done in the head, we would have no hesitation in recognising as a mental process, then that external process should be considered part of the cognitive process. Using Tetris to illustrate their point, they argue that rotating the Tetris blocks mentally or by pressing a button to physically rotate them are equivalent cognitive processes. This theory extends the traditional view that cognition is brain-bound, suggesting that the mind can encompass tools, environments, and other individuals.

In Supersizing the Mind: Embodiment, Action, and Cognitive Extension (2008), he explores the extended mind, where our mental events and consciousness reach out into the world, using technologies from simple tools such as pens through to smartphones and artificial intelligence. This was followed up by the co-authored Surfing Uncertainty: Prediction, Action, and the Embodied Mind (2015) which takes the embodied mind idea further with his theory of predictive processing and how the brain deals with uncertainty through prediction.

The extended mind is the idea that mind, body, and tools are all part of a coupled system that forms the basis of cognitive processes, that human cognition is deeply intertwined with the physical body and the external environment, suggesting that tools and technologies become integral parts of our mental processes. This perspective has implications for how we understand both natural intelligence and artificial intelligence.

Predictive engine

Regarding the mind as a predictive engine, Clark, along with other cognitive scientists, supports the idea of predictive processing or predictive coding. The brain engages in a sort of controlled hallucination, constantly making predictions about incoming sensory information and updating its models of the world based on the prediction errors it receives. More than this, the mind is an anticipatory mechanism, always trying to reduce the difference between what it expects and what it experiences. This view of the mind aligns with Karl Friston's work on the free-energy principle, which Clark has written about and integrated into his own work.

These theories see the brain from the inside out, rather than outside in. The brain is constantly running a model of the world and actively monitoring the world with weighted predictions with attention mechanisms.

Artificial Intelligence

Clark's work often intersects with discussions on AI, as he reflects on how technology continuously shapes and extends our cognitive repertoire. He sees AI as potentially becoming part of the human cognitive system, extending our minds beyond their natural boundaries and suggests that the integration of AI into human cognition can enhance and transform our mental capabilities.


The extended mind has implications in teaching and learning. It accepts the fact that technology is part of the way we can both teach and learn with the added provision that it is how we actually operate in the real world. We are extended minds and need to see ourselves in that light. Artificial intelligence, especially generative AI, which uses our cultural heritage as data, is in a sense ‘us’. 

We no longer see technology, such as drawing and writing instruments, that have been with is for around 50,000 years, as technology. The pen and pencil are invisible as technology. And so it will be with artificial intelligence. It will be folded into what it is to be human.

Neurodiversity and psychosis

One consequence of the predictive, computational model of the brain is its ability to explain neurodiversity, autism, mental illnesses such as schizophrenia and hallucinogenic drug experiences. Predictive processing models suggest that psychosis can arise from disruptions in the brain's ability to accurately predict sensory input, leading to hallucinations or delusions. Similarly, the effects of drugs on perception and cognition can be understood as alterations in the brain's predictive models, changing the way it interprets sensory information or the level of confidence it has in its predictions. Dreams can also be seen as a state where the brain generates internal predictions absent of external sensory input. This could explain the often bizarre and illogical nature of dreams, as the brain's predictive models operate without the grounding of real-world sensory data.

He also sees mediation as a form of control over these predictive engines. Our brains turn incoming data upside down, get confused when faced with alternatives and do have do deal with prediction error signals.


Just because a tool is crucial for a cognitive process, it doesn't mean it becomes part of the mind. The distinction between being merely a useful aid and actually constituting part of the mind is controversial. There are also questions about how far this extension goes and what qualifies as a part of the extended mind. 


Clark, A., (2023). The experience machine: How our minds predict and shape reality. Pantheon.

Clark, A. (2008). Supersizing the mind: Embodiment, action, and cognitive extension. New York, NY: Oxford University Press.

Clark, A. (2003). Natural-born cyborgs: Minds, technologies, and the future of human intelligence. Oxford, England: Oxford University Press.

Clark, A., & Chalmers, D. (1998). The Extended Mind. Analysis, 58, 7-19.

Tenenbaum on one or few shot systems and common sense core

Josh Tenenbaum is a cognitive scientist and professor of Cognitive Science and Computation at the Massachusetts Institute of Technology (MIT) is known for his work in the field of mathematical psychology, Bayesian cognitive science, human cognition and artificial intelligence, with a focus on understanding the nature of intelligence and creating computational models that mirror this understanding to build intelligent machines. 

He is a researcher in the Center for Brains, Minds, and Machines (CBMM), a multi-institutional collaboration that seeks to understand intelligence and develop more human-like artificial intelligence.

His work is interdisciplinary, crossing boundaries between cognitive science, computer science, and artificial intelligence, contributing both to theoretical insights about human cognition and practical applications in AI.

Computational mind

Tenenbaum's theories build upon the idea of the human mind as a computational system. One of his contributions is in the area of probabilistic models of cognition, where he suggests that the human brain operates as an inference engine that constantly predicts, analyses, and updates the world's state based on incomplete information, importantly and practically, a process that can be modeled and replicated in AI.

Bayesian models

Like Karl Friston, he uses Bayesian models and aims to build machines that can learn about the world in more human-like ways, including understanding physics, psychology and learning new concepts from limited data. AI systems that use Bayesian models belong to a category of machine learning that incorporates Bayesian statistics to infer probability distributions and make predictions. These models are known for their ability to learn from limited data by incorporating prior knowledge into the learning process. This is in contrast with many current machine learning systems that often rely on large datasets to learn effectively. Bayesian approaches can be very powerful in situations where data is scarce or when incorporating expert knowledge is crucial.

He has also proposed a ‘Bayesian Program Learning’ framework, where instead of just learning patterns in data, machines would be able to learn the algorithms themselves that generate the data.

One or few shot systems

Humans are exceptionally good at learning quickly and efficiently. For instance, a child can often recognise a new animal from just one picture and then identify it in various contexts. This contrasts with traditional machine learning systems, which typically require thousands of examples of such animals to achieve a similar level of recognition. This is because we humans bring a wealth of prior knowledge and context to new learning situations, which allows us to make inferences from sparse data. Human learning is also highly adaptable. We can apply what we have learned in one domain to a completely different domain. Understanding how to incorporate prior knowledge into AI systems can provide insights into cognitive processes like reasoning, generalisation, and conceptual learning.

AI systems that can learn from a few examples also demonstrate a similar kind of adaptability, hinting at the underlying flexibility of human cognition. A good example is Google's Alpa software. These systems in AI often use techniques like transfer learning, where a model trained on one task is adapted to another related task with minimal additional data, or meta-learning, where the model learns about the learning process itself during training to better adapt to new tasks with limited data.

In AI these one-shot or few-shot learning systems are designed to learn information or recognise patterns from a very limited amount of data – typically only one or a very few examples and aim to mimic a more human-like ability to learn quickly and efficiently from a minimal amount of information.

By creating and studying AI systems capable of one-shot and few-shot learning, researchers can test hypotheses about human cognition and develop computational models that may reflect aspects of human thought processes. It is a two-way street; not only does our understanding of the brain inform the design of AI systems, but the behaviour of AI systems can also provide clues about the principles of human intelligence.

Common sense core

Tenenbaum explores the idea of building machines that learn and think, like humans, through the development of more human-like artificial intelligence. A key concept is the creation of a ‘common sense core’ for AI, which would allow machines to use intuitive physics and psychology to understand the world in a way that is similar to how a young child learns.

Even in early childhood, we seem to learn very quickly about the physical and social world. This includes basic knowledge about objects, agents and the way they interact within the world's physical laws. We also have a knowledge that other people have minds, knowing they have beliefs, desires and goals that drive their actions.

Tenenbaum's work claims that this core knowledge is integral to human cognitive development and is something that artificial intelligence lacks. The goal of the common sense core idea in AI research is to build machines with a similar foundational understanding so that they can navigate and learn from the world in a way that is analogous to how children learn and operate in the world..

Importantly, this core knowledge would not have to be explicitly programmed for every possible scenario, instead, the machine would use it as a basis to learn and infer a wide array of situations and tasks, just like a child who leverages intuitive physics to understand that a ball thrown in the air will come down without needing to learn the exact equations of motion each time.

In computational terms, this might involve creating initial models in AI systems that reflect these basic understandings, which can then be refined through experience and learning. By incorporating this core knowledge, AI could perform better in a variety of tasks that require an understanding of the everyday, physical world, and could interact more naturally with humans by having a more aligned perspective on how the world operates. This the line taken by many other AI researchers, most notably, Yann LeCun.


This line of inquiry in connectionist thought deals with what some would see as innate qualities of the mind, dispositions which we are born with, as opposed to pure sensory empiricism, where everything is seen as the result of incoming data. The existence and reading of other minds, along with an understanding of the world and the behaviour of things in that world takes us beyond the training of large models into the replication of intelligent models that come with these capabilities. Development of AI with a common sense core could be transformative, enabling more robust and adaptable AI systems capable of reasoning and learning from minimal data, much like humans.


Critics argue that these models, while elegant, may oversimplify the complexity and variability of human cognition. The challenge in to ensure that these models capture the nuances of real-world human thinking and learning. Some of his models, particularly those addressing high-level cognitive functions, might face questions about their scope of applicability. Critics could argue that these models need to be tested across various cognitive tasks to truly assess their versatility. To be fair this work is in its infancy and he recognises the need for wider testing. Tenenbaum's work is highly interdisciplinary, spanning areas like psychology, computer science, and neuroscience. While this is a strength, it also presents challenges in integrating methods and theories from these diverse fields without oversimplifying or losing critical aspects unique to each discipline.


For those in the AI field or interested in cognitive science and machine learning, Tenenbaum's work opens up ideas and insights into the intricacies of human cognition and its application to developing intelligent machines. His publications not only offer rich theoretical frameworks but also practical implementations and experiments that push the boundaries of our understanding of artificial intelligence.

Friston on the Bayesian mind

Karl Friston is a Professor of Imaging Neuroscience and Wellcome Principal Research Fellow, a neuroscientist known for his work on brain imaging and theoretical neuroscience. As one of the most cited neuroscientists, having invented statistical parametric mapping, an international standard for analysing imaging data, he is also the recipient of many awards.

He is also known for his work in theoretical neuroscience, looking at AI and the brain, with theories around the free energy principle and active inference. In 2022 he joined the US cognitive computing company VERSES.

Free Energy Principle

In The free-energy principle: a unified brain theory? (2010) he defined the brain as an organ that tries to minimise ‘free energy’. This is an information theory ‘measure’ of the surprise or prediction errors encountered. By minimising free energy, the brain tries to reduce the difference between its predicted model of the world and incoming sensory inputs. In simpler terms, our brains constantly try to align their understanding or model of the world with the actual information received through our senses. When there is a mismatch between what the brain expects and what it perceives, it adjusts its expectations to reduce this gap. The theory is formulated in mathematics.

Imagine the brain being a captain of a ship sailing through fog. The captain (the brain) tries to minimise surprises or unexpected changes (free energy) by predicting the route ahead as accurately as possible. When the actual sea conditions differ from the captain's predictions, the captain adjusts the course to reduce these surprises. Similarly, the brain constantly updates its understanding to minimise the difference between its predictions and what it actually experiences.

Active Inference

Building on the ‘free energy’ idea, active inference holds that the brain actively samples and acts on the world to gather sensory data that confirms its predictive models, rather than just passively perceiving. Active inference is our constant attempts to try to reduce uncertainty about the world and readjusting between what our brains expect to happen and what actually happens. You may want to return serve in tennis, based on your prediction of where the ball will be relative to your body and racket. As you get better you readjust so that your predictions match the better outcome. This constant loop of predicting, checking, and acting is what Friston's active inference theory is all about.

Inference in conscious perception is an idea that goes back to 1867 with Hermann von Helmholz and his Treatise on Physiological Objects. Our brains are networks of inferences where we solve perceptual, geometric and visual cues, such as we find in faces and reading, similarly in listening and understanding speech. The brain is a connected web of super-fast inferencing, settling on solutions or best-fits, constantly optimising in the light of new data and providing continuity in consciousness.

Bayesian Brain Hypothesis

Friston views perception and cognition as fundamentally Bayesian processes. A Bayesian model for the brain sees the brain process information as a Bayesian inference machine that continuously updates its beliefs about the world based on incoming sensory information and prior knowledge or expectations. The brain has pre-existing beliefs or knowledge about how the world works. These beliefs are formed from past experiences and learning. In Bayesian terms, these are called ‘priors’ and as we interact with the world, our senses provide us with new information. This is the data that the brain uses to update its beliefs. The brain evaluates how likely the new sensory information is, given the current beliefs or hypotheses about the world, then combines the new information with its prior beliefs to form updated beliefs. This process is mathematically described as calculating the ‘posterior’ probability, which integrates the likelihood of the new information with the prior probability. The brain makes further predictions about future sensory input based on these updated beliefs. If the new sensory input does not align with predictions, the brain experiences a prediction error, which it then uses to further update and refine its beliefs.

This Bayesian approach to understanding the brain is part of a broader movement in cognitive science and neuroscience that views the brain as an active inference machine, rather than a passive information processor. It has implications for understanding perception, decision-making, and even psychopathologies. For example, hallucinations in schizophrenia can be thought of as a case where the brain gives too much weight to its priors, leading to distorted perceptions of reality.

He includes another level of theorising around Hierarchical Predictive Coding, namely coding across cortical layers of the brain, with higher levels generating top-down predictions that are matched against bottom-up sensory evidence. In computing, there are processes that are analogous to Hierarchical Predictive Coding. For example, machine learning models, particularly in deep learning, often use a hierarchical structure where layers of processing are used to build up increasingly complex representations of data. Neural networks, for instance, can be trained using backpropagation, which is an algorithm that minimises the prediction error across the layers of the network.

Unified Brain Theory

Friston aims for an overarching theory where processes like perception, action, reasoning and learning all follow from the same core principles of prediction error minimisation. In essence, Friston sees the brain as a hierarchical prediction machine constantly trying to match its predictive models of the world with sensory inputs through Bayesian belief, updating across multiple neural levels. This perspective aims to provide a unified account of different cognitive functions based on a computational model of the brain.


Karl Friston's theories, including the Free Energy Principle and Active Inference, although influential, are complex mathematical formulations, even for experts to understand. The hypothesis is also very broad, attempting to explain all brain processes and has been criticised for failing to explain the sheer variety of neural and cognitive processes, Some also claim the theory does not actually map onto the actual physiology of the brain. The main criticism is that it is reductionist, reducing all brain functions to mathematical models that ignore the richness of internal subjective experience, language and culture.


Despite these critiques, Friston's work remains a unique cornerstone of theoretical neuroscience, stimulating much research and discussion in the field, especially in the light of the success of AI. The debate around his theories reflects the broader challenges of understanding the intricate workings of the brain, enlivened by advances in smart technology.


The convergence of cognitive science, neuroscience, brain scanning, mathematics and informed by successes in AI has stimulated ideas around the connected network of the brain an incredibly powerful computer. It is not that the brain does exactly what multimodal models do in terms of being trained, fine-tuned, even altered by other humans correcting our models. What we are seeing are more like strong clues as to how a computational brain could work that matches the experimental evidence in perception and action. Theory and experimental evidence is in its infancy but it may just allow us to escape from the Cartesian idea of the self and mind, defined in terms of fixed cognitive concepts, modules and fictitious unconscious entities.


Friston, K. (2010). "The free-energy principle: a unified brain theory?" Nature Reviews Neuroscience, 11(2), 127-138.

Dennett, AI and the computational mind

Daniel Dennett is an American philosopher who studied under Gilbert Ryle, the famous anti-dualist philosopher, at the University of Oxford and became the co-director of the Center for Cognitive Studies and a Professor of Philosophy at Tufts University. 

A renowned philosopher but also a polymath, with a deep understanding of psychology, science and maths, he is known for his controversial philosophy of consciousness and philosophy of mind, especially in relation to evolutionary theory and cognitive science. He has played a formative role in connectionism, as he has helped create this new multidisciplinary field of psychology that includes philosophy, psychology, evolution, neuroscience, computer science artificial intelligence and language. 


In Darwin's Dangerous Idea (1995) he explains the evolution of Consciousness arguing that, just as physical traits are subject to evolutionary pressures, so too are mental faculties like consciousness, which have evolved due to their adaptive advantages. This is a hallmark of his work, a commitment to evolution as a foundational, naturalistic theory that shaped the mind as opposed to dualistic views of the mind. His criticism of Cartesian dualism, the idea that mind and body are fundamentally different substances, argues for a more materialistic understanding of consciousness, suggesting that what we call the mind is simply a product of physical processes in the brain with an emphasis on the computation theory of mind.

Computational theory of the mind

Dennett advocates for a form of functionalism, suggesting that mental states are defined by their functional roles, not by their internal constitution. In Consciousness Explained (1991) he argues for a form of functionalism, a model of consciousness that involves multiple parallel processes and interactions which can be likened to computational processes. 

It is vital, he believes, that we understand the brain's information processing mechanisms to comprehend consciousness and cognition. This is not to to see the brain as a computer a reductive fashion. We need more complex view, recognising the intricacies and unique aspects of human cognition that go beyond traditional computational models. He integrates insights from evolutionary biology, cognitive science, and artificial intelligence to explore these ideas. His emphasis is on the need for a comprehensive and interdisciplinary approach to studying consciousness and cognition.


Again, in Consciousness Explained (1991), where he offers a detailed theory of consciousness with the idea, going back to Hume and others, that a central narrative or a unified ‘self’ is an illusion created by the brain. Conscious experiences are constructed by the brain out of a variety of sensory inputs and mental processes that occur simultaneously and are then retrospectively made sense of as a coherent narrative.

His ‘Multiple Drafts’ model of consciousness, which is an alternative to the traditional idea of a Cartesian Theatre, sees multiple and parallel perceptual experiences unified into a single stream of consciousness. Dennett argues that various events, called ‘drafts’, happen in different places in the brain simultaneously and without any central place where can be said to be a self. There is no central theatre where consciousness comes together; rather, our brains create a coherent narrative from these overlapping drafts.

Bayesian brain

He adds to this ‘multiple drafts’ model of consciousness, a computational model of the mind and intentional stance. His vision, which has gained lots of recent traction in cognitive science, is that the brain uses Bayesian hierarchical coding, a prediction machine, constantly modelling forward. 

He sees this as the cause of dreams and hallucinations – random and arbitrary attempts at Bayesian prediction, an interesting species of the computational model of the brain that explains why the networked structure of the brain has been a productive, intuitive source of inspiration for AI, especially neural networks. It also explains why there has been interest in reversing that trend so that our knowledge of cognition can learn from computer science and artificial intelligence.


Inspired by Richard Dawkins' concept of the meme, Dennett has explored the idea of memetics, which views ideas, beliefs, and cultural phenomena as replicators similar to genes, subject to evolutionary pressures. Language plays a significant role in consciousness for Dennett and he suggests that our ability to report on our experiences is not just a by-product of consciousness but an integral part of its operation. 

He examines cultural evolution, such as religion, as the invasion or infection of the brain by memes, primarily words, and that these memes operate in a sort of Bayesian marketplace, without a single soul or executive function. These informational memes, like Darwinian evolution, show fitness in the sense of being advantageous to themselves. That brings us back to the ethical considerations around AI. He surfaces the contribution of Baldwin as an evolutionary theorist who saw 'learning' as an evolutionary accelerator.

Artificial Intelligence

His book Bacteria to Bach and Back (2017) is essential reading for those who want to see a synthesis of human evolution and AI, and its consequences for the ethics of AI. Dennett has the background depth in philosophy to understand the exaggerations and limits of AI and its consequences for ethics. He also grounds his views in biology and ourselves as humans, as both benchmark and progenitor of AI.

Just as the Darwinian evolution of life over nearly 4 billion years has been ‘competence without comprehension’ the result of blind design, what Dawkins called the ‘blind watchmaker’, so cultural evolution and AI is often competence without comprehension. This is the context of his views on AI.

Watson may have won Jeopardy but it didn’t know it had won. This basic concept of competence without comprehension is something that has to be understood to counter the exaggeration and anthropomorphism around in the subject, especially in the ethics debate. We have all sorts of memes in our brains but it is not clear that we know why they are there. Similarly with AI,

As he rightly says, we make children without actually understanding the entirety of the process, so it is with generative technology. Almost all AI is parasitic on human achievements, corpuses of text, images, video, music, maths and so on. We already trust systems that are way more competent than we humans and so we should. His call is for us to keep an eye on the boundaries between mind and machine, as we have a tendency to over-estimate the 'comprehension' of the machines, way beyond their actual competence. We all too readily read intentionality, comprehension, even consciousness into technology when it is completely absent.


His hope is that machines will open up “notorious pedagogical bottlenecks” even “imagination prostheses” working with and for us to solve big problems. We must recognise that the future is only partly, yet largely, in our control. Let our artificial intelligences depend on us even as we become “more warily dependent on them.”

AI, for Dennett, is a powerful cognitive tool that can augment human abilities, especially in the context of education, as it could be used to enhance teaching methods, personalise learning experiences and help students and educators identify and overcome learning challenges more efficiently. Also, he often explores how understanding AI can lead to better insights into human cognition. This perspective implies that advances in AI could lead to improved educational strategies by providing a deeper understanding of how we learn and process information.

His interdisciplinary approach to AI, combining philosophy, cognitive science, and biology, could influence educational models, encouraging interdisciplinary learning and the integration of AI into various fields of study.


Some feel he too easily that he explains consciousness away too easily, taking too reductive an approach  to the rich, subjective experiences and emnergent properties of cognitive processes, cognition and agency. Dennett’s comparisons between AI and human cognition are sometimes seen as overstretched and too speculative. They argue that equating human cognitive processes with computational models may ignore the unique, non-computational aspects of human thought and consciousness.


Dennett's theories are influential because they blend philosophical inquiry with empirical science, providing a framework for understanding the mind and consciousness in terms of physical processes and evolutionary biology. His work has contributed significantly to debates in philosophy of mind, cognitive science, and artificial intelligence.


Dennett, D. C. (1969). Content and consciousness. London: Routledge & Kegan Paul.

Dennett, D. C. (1978). Brainstorms: Philosophical essays on mind and psychology. Montgomery, VT: Bradford Books.

Dennett, D. C. (1984). Elbow room: The varieties of free will worth wanting. Cambridge, MA: MIT Press.

Dennett, D. C. (1987). The intentional stance. Cambridge, MA: MIT Press.

Dennett, D. C. (1991). Consciousness explained. Boston, MA: Little, Brown and Co.

Dennett, D. C. (1995). Darwin's dangerous idea: Evolution and the meanings of life. New York, NY: Simon & Schuster.

Dennett, D.C., (2017). From bacteria to Bach and back: The evolution of minds. WW Norton & Company.

Strawson on the storied self

Galen Strawson is the philosopher son of another famous philosopher P.F Strawson. He has held academic positions at several universities, including the University of Oxford, the University of Reading and currently at the University of Texas at Austin.

His book The Evident Connexion (2011) expands upon Hume's notion of personal identity, although he has wider interests in free will and determinism and panpsychism.

Personal identity

Strawson is well-known for his argument against the narrative self and for his defense of ‘episodic’ selves, those who do not see their lives as a story or narrative. He also engages with the concept of panpsychism, the idea that consciousness is a fundamental aspect of all things, and has critiqued materialist views of consciousness.

In Selves (2009) he argues strongly against the orthodox idea of the self as a continuously existing entity. This challenges mainstream positions on personal identity and the nature of consciousness, as he is critical of what he calls the ‘Narrative’ view of the self, which claims that people see themselves or their lives as a narrative or story and that this is essential to a well-lived life and a sense of self. He counters this with what he calls the ‘Episodic’ view, where individuals do not see their lives as a coherent narrative and do not have any strong sense of an ongoing self. It is not necessary to see oneself in narrative terms, to have a rich and fully human existence, indeed this may limit your exploration of ways of living and beliefs.

The self, for Strawson, is a complex process, not an entity and needs to be understood as a series of connected mental events and states, but without any persistent, underlying essence that remains the same over time.

As for consciousness, despite being a physicalist even on mental states, he is also a panpsychist, seeing consciousness as a fundamental feature of the universe, present even at the atomic level. This view is very different from traditional views of consciousness as emerging solely from complex interactions within the human brain. What he does add is the idea of self-intimacy, that sense of immediate, experiential, internal knowledge one has.

Critique and discussion

His panpsychism has been criticised as being speculative and bring in a form of dualism into his position. His denial of the storied-self has also been seen as diminishing personal responsibility and a personal sense of meaning in life. Yet his negation of the storied self in favour of a more open sense of personal identity opens up possibilities for seeing life not as limited by internal storytelling but freed from its fictions and constraints.


Strawson’s work encourages us to reexamine our assumptions about our continuous identity over time, the importance of narrative in our lives, and the nature of consciousness itself. His arguments have significant implications for various areas of philosophy, including metaphysics, epistemology, and ethics.


Strawson, G. (2011). The Evident Connexion: Hume on Personal Identity. Oxford: Oxford University Press.

Strawson, G. (2009). Selves: An Essay in Revisionary Metaphysics. Oxford: Oxford University Press.


Parfit on brain states and personal identity

Derek Parfit (1942-2017) was a British philosopher who specialized in personal identity, rationality, and ethics. A professor at Oxford he is best known for his paring back of personal identity in the Humean tradition, denying any core subject, self or soul.

He was analytic, believing in the power of reason but also thought experiments to expose weaknesses and strengths around personal identity and ethical issues. His work continues to be hugely influential in the fields of philosophy and ethics.

Personal Identity

His most famous work is Reasons and Persons (1984) where he challenges conventional views of personal identity and discusses the implications of such views for ethics and rationality.

The single, continuous self is an illusion, what matters is psychological continuity and connectedness not personal identity. Evolved for survival, we rely on memory, personality, and intention, not any sense of the moral self. He holds that that personal identity comes down to particular facts about brain states, psychological continuity, and bodily continuity. There is no need for any additional ‘subject’ as self and that speculation beyond these facts is pointless and fictional.


This he believes has huge consequences in terms of ethics and in On What Matters () where a complex mix of reasoned ethics lifts us out of mere feelings and emotions. This view of personal identity breaks down the sovereign self into a more social and shared view of ethics.

He has a Triple Theory, where Kant, consequentialism, and contractualism come together to support a common set of ethical standards. His Kantian Ethics saw certain rule-based moral laws that we must follow based on the Kantian idea of universalizability. Then there is his rule consequentialism, where we agree to follow rules that are in our best interest in terms of consequences. This allows for the consideration of the consequences of actions, but within the constraints of rule-following. Finally, there is contractualism the idea that something is right or wrong for informed, unforced general agreement.

Critique and consequences

When personal identity is not seen as singular and necessary for survival it raises challenges around personal identity, responsibility and institutions that rely on that concept. It also raises questions about the status of psychological continuity. 

One can see how this idea, when applied to AI leads to us confusing AI by reading subjects, or selves into the machine, ghosts in the machine. When that goes, as it does in Large Language Models, then we have a more diffuse entity with less individualistic outputs. Without the connectedness and continuity of being a person, can AI have a sense of being in the world?


Parfit’s philosophical work has had a substantial impact on fields like ethics, philosophy of mind. His work raises questions in AI around what being human is and the consequences of a shallower, brain-process orientated sense of identity. He has been influential on theorists who see the brain He did not propose a computational theory of the brain in the sense that cognitive scientists or computational neuroscientists might buy laid the ground for later theorists to layer this on to his ideas.


Parfit, D. (2011). On what matters (Vols. 1 & 2). Oxford: Oxford University Press.

Parfit, D. (1997). Equality and priority. The Lindley Lecture. Lawrence: University of Kansas, Department of Philosophy.

Parfit, D. (1984). Reasons and persons. Oxford: Clarendon Press.


Hume as psychologist and on learning

David Hume (1711-1776), the Scottish Enlightenment philosopher, and historian is regarded by many as the greatest English-speaking philosopher.

He attended the University of Edinburgh at the unusually young age of twelve and left without graduating producing seminal texts in Western philosophy; A Treatise of Human Nature (1739-40), Enquiries concerning Human Understanding (1748), and Enquiries concerning the Principles of Morals (1751). 

These books focused on developing a naturalistic science of man that examined the psychological basis of human nature. His empiricism, theory of mind, personal identity, association of ideas and analysis of emotions and morality, along with scepticism of religion are foundational in philosophy and psychology.


Hume’s insights are precursors to several modern psychological concepts. As a pure empiricist, he believed that all human knowledge arises from sensory experience, through 'impressions' and then 'ideas', which can be linked to how we process and interpret information, a cornerstone in cognitive psychology. His principle of association, where one idea naturally leads to another influenced later psychological theories of how thoughts and memories are interconnected.

He also studied the role of emotions in thought and moral judgment, which predates the now rich field of moral psychology. Moral distinctions for Hume are derived from feelings of pleasure and pain rather than intellectual reasoning, highlighting the importance of emotions in human behaviour and decision-making.

His scepticism about our ability to truly know causes and effects along with a careful analysis of inductive thinking has implications for how we understand and study the limitations of human behaviour and mental processes. This led him to question the idea of a permanent ‘self’, proposing instead that the self is a bundle of perceptions that change over time. This anticipates discussions in psychological fields about the nature and stability of the self-concept and identity.

Personal identity

It has a long pedigree, going back to Hume’s destruction of the Cartesian idea of the mind as separate from the body.

“I never can catch myself at any time without a perception, and never can observe anything but the perception. When my perceptions are removed for any time, as by sound sleep; so long am I insensible of myself, and may truly be said not to exist. Nothing but a bundle or collection of different perceptions, which succeed each other with an inconceivable rapidity, and are in a perpetual flux and movement. Thus we feign the continued existence of the perceptions of our senses, to remove the interruption; and run into the notion of a soul, and self, and substance, to disguise the variation.”

This was an astounding moment in philosophy and psychology, as it demolished dualism and most forms of established religion, opening up the mind up to study as something more examinable and intimately related to perception.

Hume’s idea that perception is primary and that we can never escape its presence, this bundle of perceptions, that constantly change, except when consciousness is lost in sleep. He focuses on the actual flow of experience, in the moment.

“I never can catch myself at any time without a perception, and never can observe anything but the perception. When my perceptions are removed for any time, as by sound sleep; so long am I insensible of myself, and may truly be said not to exist. nothing but a bundle or collection of different perceptions, which succeed each other with an inconceivable rapidity, and are in a perpetual flux and movement.”

And so Hume rejects the idea of any cognitive subject beyond our perceptions, any underlying entity:

“Thus we feign the continued existence of the perceptions of our senses, to remove the interruption; and run into the notion of a soul, and self, and substance, to disguise the variation.”

There are many who have been inspired by this notion of personal identity as perceptual, fragmented and a process, rather than an identifiable essential self. Others have explored exactly what this perceptual process is, finding that it is not simple one of input but a complex predictive and anticipatory process.

Post-Darwinian theorists like Derek Parfitt and Galen Strawson built on Hume’s ideas to consolidate this idea of the brain and mind, defined as memories, personality and intentions. All we have are brain states, psychological and bodily continuity and need no additional ideas of subject, self or soul. Galen Strawson even dismisses the idea of the ‘storied-self’ seeing personal identity as episodic.

Daniel Dennett and goes further seeing the brain as evolved organ, with a fully intentional stance, operating through a predictive engine (AI) to produce changing drafts of consciousness, adding the use of language and memetics. Nick Chater argues for a flat mind, without the fictions of unconscious entities. Here we see a theory emerging that is sensitive to Bayesian ideas in AI and the success of Generative AI

In summary, Hume's empirical approach and skepticism about metaphysical explanations of the mind, along with his insights into human understanding and the self, resonate with these later thinkers, especially as AI begins to inform us about the mind.


Having never graduated and rejected by Edinburgh University for a Professorship, despite his reputation as the greatest living philosopher he was suspicious of academic teaching, famously stating in a letter to a friend “There is nothing to be learnt from a Professor, which is not to be met with in Books.” His suspicion of abstract religious and metaphysical speculation, divorced from empirical evidence, was also his hallmark.

“When we run over libraries, persuaded of these principles, what havoc must we make? If we take in our hand any volume of divinity or school metaphysics, for instance, let us ask, Does it contain any abstract reasoning concerning quantity or number? No. Does it contain any experimental reasoning concerning matter of fact and existence? No. Commit it then to the flames, for it can contain nothing but sophistry and illusion.”

This appeal for a more tempered qualitative and sceptical approach to philosophy and psychology was to have a profound influence on post-Enlightenment thought and science.


He is keen to assert the study of human nature and the mind as both the alpha and omega of his inquiries and point out the foolishness of stories, for example about miracles, that spread and take root in uneducated populations. In other words, his epistemology is entirely in line with the idea that the mind is a dynamic and fleeting phenomenon that evades capture through reflection and a general scepticism around abstract reason on the self, causality and induction.


Hume D. (1739). A Treatise of Human Nature. Of Personal Identity Selection from Book I, Part 4, Section 6.

Hume D. (1748) An Enquiry concerning Human Understanding, edited by Tom. L. Beauchamp, Oxford: Clarendon Press, 1999.

Hume D. (1751) An Enquiry concerning the Principles of Morals, edited by Tom L. Beauchamp, Oxford: Clarendon Press, 1998.

Thursday, March 28, 2024

Chater - The Flat Mind

Nick Chater is a Professor of Behavioural Science at Warwick Business School where he leads the Behavioural Science group, one of the largest in Europe. He is also known for his role as a scientist-in-residence on the Radio 4 series ‘The Human Zoo’. 

His work in the field is extensive, including co-authoring books and numerous papers on the human mind and rationality, challenging the whole theoretical basis of psychology and linguistics with ideas around the flat mind and language as a social construct.

The Flat Mind

In The Flat Mind (2018) starts with a quote from Dennett and presents a thought-provoking theory about human cognition and challenges the conventional belief in the depth and complexity of human thought, arguing instead, that our minds operate on a much more superficial level than traditionally thought. According to his theory, human thinking relies heavily on immediate perceptions and the present context, rather than on deep, abstract reasoning about the world. The mind is not a collection of specialised modules or faculties for different cognitive functions. Instead, he argues the mind is ‘flat’ and there are no true cognitive modules, just patterns of activation over a common coding base.

Chater proposes that rather than processing and manipulating detailed internal knowledge, our minds interpret and react to current sensory data. This approach suggests that much of our cognitive processing is less about drawing from a rich, detailed internal database of knowledge and more about improvising responses based on immediate environmental cues.

The Flat Mind shifts the focus from the idea of the mind as a deep, introspective processor to a more surface-level, reactive entity. This concept has implications for our understanding of memory, reasoning, and decision-making, suggesting that these processes are more context-dependent and less introspective than previously believed.

An important point is that the mind does appear to both read and speak one word at a time and perception has limited visual acuity and colour perception. In addition, he makes the even stronerg claim that the mind is a hoax which we play on ourselves.

Bayesian Brain Hypothesis

As we don't have unlimited cognitive resources, our decision-making exhibits ‘bounded rationality’ using heuristics and rules of thumb rather than optimal rationality and so questions the idea that cognition involves constructing rich mental representations or models of the world. Cognition emerges from simple processes operating over compressed sensory inputs. The brain essentially performs Bayesian inferences or similar probabilistic calculations to interpret sensory data and make decisions, continually updating beliefs based on new evidence. A key idea is that, in the absence of mental representations, human perception and cognition is geared towards efficiently compressing and encoding sensory information into simple codes.

In essence, Chater believes cognition arises from Bayesian probability updating over compressed sensory inputs by a flat, general-purpose mind, not specialised modules or rich mental representations. His perspective challenges many traditional assumptions in cognitive science. Visual perception is the most important of the senses and he sees thinking as largely an extension of perception.

The Language Game

In The Language Game (2022), co-authored with Morten Christiansen, he presents a radical, thought-provoking exploration of how language works and its role in human society. He argues that language originated in communicative charades then develops into other more complex forms of linguistic communications and argues against the common belief that language is a complex code passed down through generations. Instead, language is a form of social negotiation that is constantly evolving. This allows our linguistic abilities develop from social interaction rather than being hard-wired into our brains through deep-seated biological evolution. The book delves into how we use words to navigate our social worlds, construct complex ideas, and solve problems together and challenges traditional linguistic theories and provides insights into the dynamic, ever-changing nature of language as a product of social processes.


He considers the rise of Large Language Models as a confirmation of his position. Like LLMs, there are no hidden depths and nothing lurks below the surface. In fact, those constructs are misconceived as stories we tell ourselves about cognition. Our imaginations tell stories and we carry this over into views about cognition, imagining that we work from deep, pre-conceived ideas and reason, when in fact we generate words in the moment and on the fly. Just as there are no hidden depths in LLMs, there are no ghosts in the machine in our minds, no Freudian psychotherapeutic phantoms like the ego, id, superego or Jungian archetypes. There is nothing of substance in our unconsciousness, as it does not exist as traditionally framed. Our imaginations see unconsciousness as a delving down to hidden depths but there are no depths, no creatures of the deep to be found. Like Hume, Parfit and Strawson, he recognises that we are storied-selves but these stories are largely imagined, narrative fictions.


Chater, N. (2018). The Flat Mind. Penguin

Chater, N. (2022). The Language Game. Penguin