Theories of Intelligence

A Brief History of Intelligence: Evolution, AI, and the Five Breakthroughs That Made Our Brains by Max Bennett

This book is awesome! It really goes back to the origin. What I find really interesting is just how much intelligence stems from simply steering a body left or right. The author charts how evolution converges on the same learning algorithms repeatedly just as it converged on eyes and other physical features. Just have a look at the table of contents:

The world before brains The birth of good and bad The origin of emotion Associating, predicting, and the dawn of learning The Cambrian explosion The evolution of temporal difference learning The problems of pattern recognition Why life got curious The first model of the world The neural dark ages Generative models and the neocortical mystery Mice in the imaginarium Model-based reinforcement learning The secret to dishwashing robots The arms race for political saavy How to model other minds Monkey hammers and self-driving cars Why rats can’t go grocery shopping The search for human uniqueness Language in the brain The perfect storm ChatGPT and the window into the mind The sixth breakthrough.

Driven by Compression Progress: A Simple Principle Explains Essential Aspects of Subjective Beauty, Novelty, Surprise, Interestingness, Attention, Curiosity, Creativity, Art, Science, Music, Jokes

By Juergen Schmidhuber on April 15th, 2009

This gets more fascinating with each read!

Daniel Wolpert: The real reason for brains

Neuroscientist Daniel Wolpert starts from a surprising premise: the brain evolved, not to think or feel, but to control movement. In this entertaining, data-rich talk he gives us a glimpse into how the brain creates the grace and agility of human motion.

He considers himself a “movement chauvinist.”

Another related video: The Moving Mind: Neuroscience, Philosophy, and Fitness | Michael Mannino

Embodied cognition - movement is necessary for the mind.

“The brain is a dynamical system, and the mind is an emergent, self-organizing, embodied, process that happens because of movement.”

Creatures which don’t move often don’t have a brain. Example of a sea squirt.

“The juvenile sea squirt wanders through the sea searching for a suitable rock or hunk of coral to cling to and make its home for life. For this task, it has a rudimentary nervous system. When it finds its spot and takes root, it doesn’t need its brain anymore, so it eats it! It’s rather like getting tenure.”

― Daniel Dennett

The brain is a complex system: self-organization, emergence, pattern formation, coordination dynamics

“The mind is a pattern of movement.” - embodied cognition

He shows a clip of Alva Noë: You Are Not Your Brain

“Looking for consciousness in the brain is like looking for the value of money in the composition of a dollar bill or trying to find dancing in the musculature of a dancer.” - Alva

“We must perceive in order to move, but we must also move in order to perceive,” James Gibson, Psychologist.

Are Brains Bayesian?

By John Horgan on January 6, 2016

Kicking things off was Joshua Tenenbaum of MIT’s brain and cognitive science program, who tries to “reverse engineer” human minds and replicate their performance in computers. As he explains on his website (which links to papers on the Bayesian brain and related topics), “bringing machine-learning algorithms closer to the capacities of human learning should lead to more powerful AI systems as well as more powerful theoretical paradigms for understanding human cognition.”

His presentation reprised his co-written 2012 paper “Bayesian just-so stories in psychology and neuroscience”, which evokes a famous complaint by biologist Stephen Jay Gould about the flimsy, ad hoc style of some evolutionary accounts of biological traits. Gould compared such explanations to “just-so stories,” fanciful tales about how the leopard got his spots and the camel his hump. In the same way, Bowers contended, Bayesian models can replicate virtually any cognitive task, given tweaking of prior assumptions and input. They are so flexible that they are immune to falsification, much like the explanations that evolutionary psychology offers for human traits.

So who wins the Bayesian-brain debate? I hate to be so predictable, but I must give the nod to Bowers, the skeptic. My coverage of brain-and-mind research over the last few decades has left me with a strong bias against alleged breakthroughs.

Imperfect Bayesian inference in visual perception

Published: April 18, 2019

Abstract:

Optimal Bayesian models have been highly successful in describing human performance on perceptual decision-making tasks, such as cue combination and visual search. However, recent studies have argued that these models are often overly flexible and therefore lack explanatory power. Moreover, there are indications that neural computation is inherently imprecise, which makes it implausible that humans would perform optimally on any non-trivial task. Here, we reconsider human performance on a visual-search task by using an approach that constrains model flexibility and tests for computational imperfections. Subjects performed a target detection task in which targets and distractors were tilted ellipses with orientations drawn from Gaussian distributions with different means. We varied the amount of overlap between these distributions to create multiple levels of external uncertainty. We also varied the level of sensory noise, by testing subjects under both short and unlimited display times. On average, empirical performance—measured as d’—fell 18.1% short of optimal performance. We found no evidence that the magnitude of this suboptimality was affected by the level of internal or external uncertainty. The data were well accounted for by a Bayesian model with imperfections in its computations. This “imperfect Bayesian” model convincingly outperformed the “flawless Bayesian” model as well as all ten heuristic models that we tested. These results suggest that perception is founded on Bayesian principles, but with suboptimalities in the implementation of these principles. The view of perception as imperfect Bayesian inference can provide a middle ground between traditional Bayesian and anti-Bayesian views.

Author summary:

The main task of perceptual systems is to make truthful inferences about the environment. The sensory input to these systems is often astonishingly imprecise, which makes human perception prone to error. Nevertheless, numerous studies have reported that humans often perform as accurately as is possible given these sensory imprecisions. This suggests that the brain makes optimal use of the sensory input and computes without error. The validity of this claim has recently been questioned for two reasons. First, it has been argued that a lot of the evidence for optimality comes from studies that used overly flexible models. Second, optimality in human perception is implausible due to limitations inherent to neural systems. In this study, we reconsider optimality in a standard visual perception task by devising a research method that addresses both concerns. In contrast to previous studies, we find clear indications of suboptimalities. Our data are best explained by a model that is based on the optimal decision strategy, but with imperfections in its execution.

Is Artificial General Intelligence a Mathematical Pattern?

Posted Mar 28, 2018

“If achieving artificial general intelligence is indeed a pattern that already exists, uncovering it involves mathematics, the science of patterns. Mathematicians look for patterns to form a conclusion, called a conjecture, and set out to support the proposition by creating a proof, or theorem.”

In sum, is AGI a mathematical pattern that can be implemented? Definitely maybe.

See Ben Goertzel posting in AGI

The humanity switch: How one gene made us brainier

3 May 2012

“When it comes to brain development, slow and steady wins the race. A single ancestral human gene that made two copies of itself may have helped the evolution of our large brains 2.5 million years ago, as our ancestors were diverging from australopithecines.

Paradoxically, it seems the effect of the extra copies was to slow down individual brain development. This allowed time for neurons to develop more and better connections with one another.

Gene duplications are rare in human history: only about 30 genes have copied themselves since we split from chimps 6 million years ago. Few have been studied, but those that have encode genes that are very exciting, says human geneticist Evan Eichler of the University of Washington in Seattle. Many are involved in brain development.”

The amazing brains and morphing skin of octopuses and other cephalopods | Roger Hanlon

Published on Jun 28, 2019

“Octopus, squid and cuttlefish – collectively known as cephalopods – have strange, massive, distributed brains. What do they do with all that neural power? Dive into the ocean with marine biologist Roger Hanlon, who shares astonishing footage of the camouflaging abilities of cephalopods, which can change their skin color and texture in a flash. Learn how their smart skin, and their ability to deploy it in sophisticated ways, could be evidence of an alternative form of intelligence – and how it could lead to breakthroughs in AI, fabrics, cosmetics and beyond.”

Jeff Hawkins: Thousand Brains Theory of Intelligence - Artificial Intelligence (AI) Podcast

“Jeff Hawkins is the founder of Redwood Center for Theoretical Neuroscience in 2002 and Numenta in 2005. In his 2004 book titled On Intelligence, and in his research before and after, he and his team have worked to reverse-engineer the neocortex and propose artificial intelligence architectures, approaches, and ideas that are inspired by the human brain. These ideas include Hierarchical Temporal Memory (HTM) from 2004 and The Thousand Brains Theory of Intelligence from 2017.”

A Framework for Intelligence and Cortical Function Based on Grid Cells in the Neocortex

On Intelligence: How a New Understanding of the Brain Will Lead to the Creation of Truly Intelligent Machines

Numenta - Advancing Machine Intelligence with Neuroscience

YouTube comment: Cleon Teunissen

40 minutes in: the view presented by Jeff Hawkins strongly reminds me of the views presented by Marvin Minsky in his 1986 book ‘The society of mind’. Highly recommended, that book.

Minsky offers a theory of psychology (both cognitive psychology and personality psychology). Minsky presents that the evidence suggests that we should think of the psyche as a multitude of cooperating agents. This society strives to reach decision by concensus, but that is not always possible.

Minsky discusses: how is our brain able to process language so fast, given how slow neurons are? Minsky suggests massive parallellism. When words enter the auditory center they spread thoughout all of the language center, and all possible associations are generated, including the most farfetched. All those associations in the overall context are compared with each other simultaneously, one that outcompetes all others makes it to the level of conscious thought. Almost always the winning assocation is in fact the correct one. All that in a fraction of a second. Minsky offers that this is why puns strike us as funny; a “wrong” meaning, that otherwise would not make it to the level of conscious thought, is suddenly justified.

What the brains of people with excellent general knowledge look like

First published: 28 July 2019

“The brains of people with excellent general knowledge are particularly efficiently wired. This was shown by neuroscientists at Ruhr-Universität Bochum and Humboldt-Universität zu Berlin using magnetic resonance imaging. “Although we can precisely measure the general knowledge of people and this wealth of knowledge is very important for an individual’s journey through life, we currently know little about the links between general knowledge and the characteristics of the brain,” says Dr. Erhan Genç from the Department of Biopsychology in Bochum. The team describes the results in the European Journal of Personality on 28 July 2019.

The researchers examined the brains of 324 men and women with a special form of magnetic resonance imaging called diffusion tensor imaging…The participants also completed a general knowledge test called the Bochum Knowledge Test, which was developed in Bochum by Dr. Rüdiger Hossiep. It is comprised of over 300 questions from various fields of knowledge such as art and architecture or biology and chemistry…The result: People with a very efficient fibre network had more general knowledge than those with less efficient structural networking.”

Abstract

Cognitive performance varies widely between individuals and is highly influenced by structural and functional properties of the brain. In the past, neuroscientific research was principally concerned with fluid intelligence, while neglecting its equally important counterpart crystallized intelligence. Crystallized intelligence is defined as the depth and breadth of knowledge and skills that are valued by one’s culture. The accumulation of crystallized intelligence is guided by information storage capacities and is likely to be reflected in an individual’s level of general knowledge. In spite of the significant role general knowledge plays for everyday life, its neural foundation largely remains unknown. In a large sample of 324 healthy individuals, we used standard magnetic resonance imaging along with functional magnetic resonance imaging and diffusion tensor imaging to examine different estimates of brain volume and brain network connectivity and assessed their predictive power with regard to both general knowledge and fluid intelligence. Our results demonstrate that an individual’s level of general knowledge is associated with structural brain network connectivity beyond any confounding effects exerted by age or sex. Moreover, we found fluid intelligence to be best predicted by cortex volume in male subjects and functional network connectivity in female subjects. Combined, these findings potentially indicate different neural architectures for information storage and information processing.

Blue Brain finds how neurons in the mouse neocortex form billions of synaptic connections

AUGUST 30, 2019

In a paper published in Nature Communications, the Blue Brain researchers have shown that the trick lies in combining these two views. By integrating data from two recent datasets—the Allen Mouse Brain Connectivity Atlas and Janelia MouseLight—the researchers identified some of the key rules that dictate which individual neurons can form connections over large distances within the neocortex. This was possible because the two datasets complemented each other in terms of entirety of the neocortex and the cellular resolution provided.