Intelligence and its dimensions

Defining intelligence and its dimensions in order to formulate my research questions

Sep 15, 2021

After publishing my first approach to the field of augmented collective intelligence, I received interesting feedback that helped me to continue my research work.

A first comment that I got was: “Interesting elaboration, but what is intelligence to start with?” It is a fair comment because I purposely avoided defining intelligence in the first place since it is not an easy concept to grasp. But reacting to the feedback, I will try to do so in this article.

One step back: defining intelligence

When defining a concept, its etymology is always a good place to start. The Latin word intelligentia comes from intelligere, a word composed of the terms intus (among) and legere (to read or choose). Therefore, etymologically intelligence is the capacity to read or understand reality, but also the capacity to choose the best alternative. Combining both ideas, we could say that intelligence is the capacity to understand reality in order to make good decisions.

Beyond its etymology, we can look at the history of the concept of intelligence to better understand its current meaning. The first attempts to define intelligence by measuring it were those of Francis Galton in his laboratory in South Kensington museum, starting in 1882. At the museum, visitors were given a battery of tests designed to measure sensory and auditory discrimination skills as well as reaction times. After Galton’s tests, other attempts to measure intelligence were the school tests developed by Binet and Simon and the measurement of the intelligence quotient (IQ) proposed by William Stern. Another advocate of a unidimensional measure of intelligence was Charles Spearman, who claimed to have found a common factor of intelligence -the G factor- in a variety of tests designed to measure different cognitive abilities. All these developments consolidated the concept that intelligence was a unidimensional and quantifiable attribute.

But criticism of this vision soon appeared, pointing out this conceptualization of intelligence as superficial, narrow and biased. Louis Leon Thurstone was one of the first researchers who focused on revising and expanding the concept of intelligence. Between 1930 and 1940, Thurstone designed several studies in which he hypothesized that the structure of intelligence was made up of a series of differentiated abilities. Using multiple factor analysis, he proposed nine primary mental abilities: inductive reasoning, deductive reasoning, practical problem solving, verbal comprehension, associative memory, spatial visualization, perceptual speed, numerical ability and verbal fluency. Building on Thurstone's work, other researchers went on to propose many more factors of intelligence, such as Guildford who postulated 120 separate abilities that would make up intelligence.

That was the common ground until a revolutionary theory emerged in the 1980s: the theory of multiple intelligences developed by Howard Gardner. Discarding the factorial view of intelligence, Gardner proposed that intelligence could not be measured by standardized instruments in IQ tests, and alternatively offered criteria not to measure it but to observe and develop it. Based on Piaget's view of intelligence as adaptation to the environment and his own research in cognitive sciences and neuropsychology, Gardner gave new relevance to the pluralistic approach to intelligence defining it as "a biopsychological potential to process information that can be activated in a cultural setting to solve problems or create products that are of value in a culture." In his theory, Gardner proposed to broaden the spectrum of human intelligence to eight capacities considered as relevant ways of successful adaptation to the environment: linguistic, logical-mathematical, musical, spatial, bodily-kinesthetic, intrapersonal, interpersonal and naturalist.

Among the intelligences proposed by Gardner, the intrapersonal and interpersonal were the focus of the authors who developed the theory of emotional intelligence. The greatest disseminator of this new theory was Daniel Goleman, for whom emotional intelligence was the capacity to recognize one's own and others' emotions and feelings and the ability to manage them. Goleman argued that IQ was a poor predictor of success in life, and then proposed other key characteristics for intelligent performance: "abilities such as being able to motivate oneself and persist in the face of frustrations; to control impulse and delay gratification; to regulate one's moods and keep distress from swamping the ability to think; to empathize and to hope." These abilities posed by Goleman as constituents of emotional intelligence expanded the traditional concept of intelligence beyond the field of classical psychology.

From this historical review we can hence conclude that intelligence is a set of abilities that allow us to process information about oneself and the environment as well as to manage one’s own and other’s feelings and motivations in order to make decisions, solve problems and create products that are of value in a culture.

To complete this review, we can take a look at some definitions of intelligence coming from fields other than psychology, namely neuroscience and artificial intelligence.

From neuroscience, Rafael Yuste offers an insightful description of the intellectual capacity of the brain: “The purpose of the brain is to predict the future. The nervous system builds a model of the world, more or less complex depending on the animal, essentially internalizing the physical world so it can be calculated and manipulated internally to predict the future in a biological form of “virtual reality”. This virtual reality is built and manipulated internally using memories from our life’s experiences and kernels of biological wisdom ingrained by evolution in our nervous system. Using this information, and updating the predictions with our senses, we try to anticipate ourselves to the future. This is the heart of being intelligent: to guess what the future will bring.”

And from the field of artificial intelligence, David Poole poses an interesting definition of an intelligent agent: “An intelligent agent does what is appropriate for its circumstances and its goal, it is flexible to changing environments and changing goals, it learns from experience, and it makes appropriate choices given perceptual limitations and finite computation.”

As a final summary, we can conclude that intelligence:

is a set of diverse abilities -ranging from verbal comprehension to logical reasoning to managing one’s own feelings, i.e. a multi-dimensional capacity
that builds on processing information about a changing environment and the adaptation of the agent to it
that operates by modeling the world and manipulating this model of reality to make predictions about the future
so as to guide the behavior of the agent to achieve some objectives that are considered of value in a culture

How does intelligence work?

Beyond this conceptual definition of intelligence, we can also review what we know about how intelligence works. And the reality is that across its different dimensions -human intelligence, collective intelligence and artificial intelligence- we still do not know much about it.

All we don’t know about the brain. Neuroscience has made great progress over the last decades thanks to the development of new technologies to record brain activity and advanced computational models to process this data. But at the same time, there is still so much we don't know about the development and functioning of the brain that we can hardly explain how the biophysical substrate of intelligence works. For example, we don’t really understand why during the initial development of the brain there is a massive neuronal die-off followed by another massive elimination of synapses, so that the final nervous system is sculpted out of an initially much larger number of neurons and connections between them. Neither do we understand the closing of critical periods during brain development: the brain shows plasticity in its neural development in critical periods during which neuronal activity instructs which synapses survive and which ones are pruned, but we don’t know why nature closes those critical periods that occur at different stages of brain development. Another example of a process of the brain that we are not able to explain is the stochastic behavior of synapses in the central nervous system: when an action potential invades a presynaptic terminal, 50% of the time it does release the neurotransmitters and the other 50% of the time it does not. In summary, the development and functioning of our brain still hides many mysteries.
Our still very basic collective intelligence. If we move to the dimension of collective intelligence, we also realize that our experience developing different forms of collective intelligence is still very limited. Following Thomas Malone’s framework to categorize different types of collective intelligence -or what he calls superminds-, we can distinguish hierarchies (where people with authority make decisions that others are required to follow), democracies (where decisions are made by voting), markets (where decisions are made by mutual agreement among trading partners) and communities (where decisions are made by informal consensus or shared norms). The reality, though, is that hierarchies are by its own nature very ineffective in capturing collective intelligence; existing democracies follow a very basic model of voting representatives every four years; markets can be very effective but only in seeking the unidimensional objective function of maximizing economic return; and communities are complex to manage and often chaotic. It is true that we are now seeing a new wave of experimentation with decentralized models of human collaboration in the crypto space, but these are still early days for this emergent ecosystem.
AI is harder than we think. Given that we still don’t really understand how human intelligence works, it may seem a bit pretentious to even talk about artificial intelligence. Put in other words, until we do understand the functioning of our own intelligence, we will hardly be able to create artificial intelligence if by it we mean an intellectual capacity at human level in all its dimensions. Part of the problem is that we use a quite misleading language that includes terms like artificial intelligence or neural networks to refer to mathematical models that are very effective in solving specific problems but have not much to do with the human intellectual processes that we are still trying to understand. It is true that these AI models are able to “self-calibrate” for a specific objective function, but to call that learning is also a misleading generalization in my view. As Melanie Mitchell explains in her paper “Why AI is harder than we think”, the field of artificial intelligence has cycled several times between periods of optimistic predictions and massive investment (AI springs) and periods of disappointment and reduced funding (AI winters) due to our limited understanding of the nature and complexity of intelligence itself. As Mitchell points out: “It turns out that, like all AI systems of the past, deep-learning systems can exhibit brittleness”, i.e. are often unable to generalize or adapt when faced with new situations (compare that with David Poole’s definition of intelligence above, stating that intelligence “is flexible to changing environments and changing goals”). The problem is that as we don’t understand how intelligence works, we tend to oversimplify the search for general artificial intelligence making hypothesis -that Mitchell describes as fallacies- such as that “narrow intelligence is on a continuum with general intelligence” or that “intelligence is all in the brain”, which may prove to be completely wrong. The practical problem comes from the fact that no one yet knows how to embed common sense in machines, and this may well not be just a question of building bigger and deeper neural networks.

In conclusion, given how far we are from really understanding how intelligence works, we should consider all research in this space as exploratory and subject to continuous review and development.

Dimensions of intelligence

Another feedback that I got from my first article was the following question: “Are the three dimensions of intelligence (human intelligence, collective intelligence and augmented intelligence) fundamental and independent?” Fundamental meaning that they are of a basic nature and independent meaning that each one of them has its own entity.

My answer is no. We can consider them as three “dimensions” or representations of intelligence but they are actually interconnected (and therefore not independent) and they are of a hierarchically different nature (so not equally fundamental).

Regarding their interdependency, it is evident that collective intelligence is dependent upon the existence of individual human intelligences. Following Ken Wilber’s model of holons (or whole/parts), we could say that collective intelligence is a holon that integrates many parts (many individual human intelligences) but at the same time transcends them to define a novel pattern of intelligence emerging from its wholeness. In other words, collective intelligence is formed by a network of human intelligences but it is also more than just the sum of its parts as it creates a new order of complexity. In a different fashion, augmented intelligence is also dependent on the use of technology augmentation tools (such as machine learning models) by human intelligence to attain specific objectives.

Even more relevantly, these three dimensions of intelligence have a different hierarchical nature. Human intelligence -as a capacity arising from human mind and consciousness- follows a hierarchical development process by which it can evolve into higher levels of cognition and agency. This is something we can clearly understand by looking at the psychological development of a human being from childhood to adulthood, passing through different stages of mind-consciousness that Wilber defined as sensoriperceptual, emotional, symbolic, conceptual, operational… up to the trascendental levels of self. On the other hand, collective intelligence does not follow a hierarchical development by itself -as far as we know- but as a result of integrating human intelligences with a consciousness in development. In other words, the hierarchical level of a collective intelligence is given by the hierarchical development of the individual human intelligences that give rise to it. This is what Frederic Laloux describes in his book Reinventing Organizations when presenting different stages of organizational development -impulsive, conformist, achievement, pluralistic…- based on the consciousness level of the members of the organization.

Now, looking at both dimensions (human intelligence and collective intelligence), we could represent human intelligence in a vertical axis of hierarchy in which, through its development, human intelligence gains comprehensiveness and agency, i.e. depth. And we could represent collective intelligence in a horizontal axis of scale in which, through its expansion, collective intelligence gains diversity and capacity, i.e. width. In the former, the development is a process of transcendence to higher levels of consciousness (or hierarchy); in the latter, the expansion is a process of integration of a larger number of individuals (at a given level of consciousness).

Finally, we could represent augmented intelligence in a third axis of skill in which the nodes in the network (at a given level of consciousness and connecting to other nodes) can improve its ability to solve certain problems by integrating technology enabled tools like deep neural networks, thus gaining accuracy and efficiency, i.e. length.

In summary, although we can consider these three dimensions of intelligence conceptually relevant, they show a clear order of priority given that hierarchy is of higher order than scale, and -in the context of this discussion- scale is of higher order than skill. In other words, the nodes of the network being able to raise their level of consciousness (or hierarchy) is of higher relevance than the network being able to extend its size (or scale) which is of higher relevance than the nodes being able to augment their abilities (or skills). As this may sound a bit conceptual, putting it in plain words it is like saying that a small group of adults has greater intelligence than a large group of babies (hierarchy vs scale) and that a large and diverse community of people has greater intelligence than a small group equipped with AI algorithms (scale vs skill).

Defining my research questions

The previous conclusion leads us to the third element of feedback that I got on my article: “For greater clarity, what are your research questions?”

This is again a fair question because I was not very specific about my research questions in my first article. The reason for this was that I wanted to follow an agile research process of exploration, learning, definition and iteration rather than setting my research questions upfront. That being said, I believe at this point I can define my initial research questions based on the conclusions of the previous section.

In practical terms, the goal of my research is helping to create better human organizations by understanding and leveraging augmented collective intelligence. This means that I seek how to create human organizations -understood as networks of human collaboration that integrate technology enabled tools- that are more effective and efficient but also more human and better integrated into their social and ecological environment. Accordingly, critical questions that arise when following the three axes of development above are:

(Hierarchy - depth) How to raise the level of consciousness of individuals in an organization? I believe that the level of consciousness of people in an organization sets the limits for their capacity to effectively collaborate and give rise to a collective intelligence, rather than just being a collection of individuals. As an example, many companies try to accomplish deep organizational transformations that in practice don’t work because their people -starting with their leadership team- are not in a high enough level of consciousness to transcend their egos and truly pursue collective goals. This implies that organizations can and should play a role fostering not only professional development but also personal growth. Some companies are already moving in this direction by offering programs that go beyond professional learning and development (like mindfulness training or psychological well-being) or focusing on new factors in their recruiting processes (such as humility and empathy).
(Scale - width) How to extend and improve the connectivity between people in an organization? Beyond the opportunity to grow the network and gain scale, the real question is how to make the network an effective system of communication, interaction and decision-making. This question pretends to look at the everlasting challenges of organizational design, culture and leadership from the point of view of network theory, including the analysis of new decentralized forms of organization emerging in crypto ecosystems.
(Skill - length) How to effectively integrate augmentation technologies into a human network? Beyond the opportunity to increase the organization’s capabilities by developing a culture of continuous learning (a very fundamental theme that I will not cover here), the question I pose is how to effectively integrate technology and data enabled tools into complex networks of human collaboration so as to extract all gains from these tools while giving newer and more human roles to the individuals in the organization and re-skilling them for these roles.

These three questions will be the focus of my research in the coming months. They touch perennial themes such as promoting collaboration between teams, designing effective organizations and adopting new technologies, but they do so from novel points of view coming from consciousness development, network theory and re-skilling, respectively.

A conscious leadership team

To conclude this article, I want to share a first look at the question of consciousness development as it is the most important by hierarchical status.

The level of consciousness of an organization is given by the level of consciousness of its individuals. But among those, the leadership team exerts a multiplying influence on the rest of the organization. In other words, the consciousness level of an organization can hardly be higher than that of its leadership team, while a rise in the consciousness of the leadership team will likely pull the rest of the organization upwards. The problem is that in some cases the members of the leadership team tend to be egocentric, as a consequence of having had a very successful career and being surrounded by flattering persons. Therefore, those leadership teams that are able to make progress as a team in transcending their egos can have an enormous impact in their organizations.

I will not review here the different stages of consciousness development and how they lead to transcend the ego, but the translation of these ideas into a leadership model is something that Bob Anderson and Bill Adams very well accomplished in their article Five Levels of Leadership. In this model, Integral Leadership -as an advanced leadership level- can be simply described as a leadership of service: “the leader becomes the servant of the whole”. This requires a high level of consciousness that manifests in basic virtues such as humility, generosity and magnanimity. Interestingly enough, these virtues have been part of the leadership models of both ancient philosophers such as Socrates and modern management experts like Jim Collins. In fact, Socrates already said that “pride divides the men, humility joins them”. And quoting Jim Collins to close: “Great leaders are differentiated from other leaders in that they have a wonderful blend of personal humility combined with extraordinary professional will. Understand that they are very ambitious; but their ambition, first and foremost, is for the company's success. They realize that the most important step they must make to become a great leader is to subjugate their ego to the company's performance”.

Note: I have published some additional thoughts on conscious leadership here

Augmented collective intelligence

Discussion about this post