In Part1 of my extended account of Process Overlap Theory I explained how general intelligence (g) is a statistical fiction that doesn’t represent any real psychological processes or neural mechanisms in the brain.
What is real in the brain when we take an IQ test or apply our intelligence are ‘domain general’ executive processes (with working memory and attention control playing the central role) located in fronto-parietal executive networks, and ‘domain specific’ processes located in visuospatial, verbal, numerical, etc, brain networks.
It is the overlapping executive processes across different domains – according to the Process Overlap Theory (article) I endorse – that actually explains general intelligence. These overlapping processes are shown by the black dots in both the verbal comprehension (Gc) and visuospatial (Gc) factors of g in the POT model shown below.
Extended Process Overlap Theory
The POT model allows for a wide array of specialized domains of cognition that can be measured as factors in different types of tests – not just the abilities measured by verbal and visuospatial tests. There is evidence for all of the different types of cognitive abilities – which we can call ‘multiple intelligences’ – in the extension of the POT model shown below:
Working memory and attention control processes that dominate in fluid intelligence (Gf) are also at work in reading/writing challenges, quantitative problems, emotional challenges, complexity in our social lives requiring social cognition, and in any number of complex sensorimotor skills (DIY skills, sports, musical instruments, how to play computer games, etc).
Performance in all of these cognitive domains correlate positively with fluid intelligence, due to the overlapping executive processes.
Fluid intelligence redefined: using executive processes (working memory and attention) to reason, problem solve, and to see new conceptual relations in any of the specific domains of cognition shown above when the cognitive challenge is novel and complex.
Matrices and number series tests are two well-known fluid intelligence tests – in the visuospatial and numerical domains. But in principle – on this account – we could use reading or writing, social cognition or sensorimotor tests to measure fluid intelligence – provided the challenges they posed were complex and novel enough.
The Fluid-Intelligence – Crystallized Intelligence (Gf-Gc) Continuum
The main idea of the well-known Gf-Gc model (Cattell & his student Horn) has been defined as the distinction between:
the ability to solve problems in novel situations, regardless of previously acquired knowledge (fluid intelligence or Gf)
And
the ability to solve problems using already acquired concepts, skills or knowledge in long-term memory (crystallized intelligence or Gc)
There is naturally a continuum between Gf and Gc. Fluid intelligence cannot function without drawing on previously acquired knowledge in long-term memory. When you figure out a number series or a matrices problem (such as shown below) you make use of concepts, rules, memories, strategies and other types of knowledge stored in your long term memory.
According to Process Overlap Theory, what you are doing when you solve a new matrix problem is to activate relevant long-term (crystallized) knowledge in working memory and use executive processes to solve it. This can be visualized in Cowan’s well-known model of working memory shown below: In this model, the Central Executive-Focus of Attention-Working Memory mechanism can be interpreted as fluid intelligence (Gf) and ‘Information in Long Term Memory’ can be interpreted as crystallized intelligence (Gc).
In the POP model depicted above , when executive processes operate on domain-specific knowledge (such as understanding how shapes rotate) and the cognitive challenges are relatively simple and/or well-known, then you are tapping relatively more domain-specific crystallized intelligence (such as Gv) and less domain-general fluid intelligence.It is a matter of degree: the more complex and novel the cognitive challenge in a specific knowledge domain, the more it requires fluid intelligence (Gf); the more simple and well-known the cognitive challenge, the more domain-specific crystallized intelligence can do the job (Gc, Gv, Gq, Grw, Gsm, Gsc, Geq).
Moreover, the executive processes engaged by fluid intelligence are domain-general: they can integrate processes from multiple domains. In the example of solving matrix problems, executive processes may activate a combination of visuospatial and verbal crystallized knowledge from long-term memory. For this reason, Gf processing is well-suited to finding relationships across domains and transferring knowledge between them. For this reason, analogical reasoning tests are good measures of fluid intelligence.
Multiple Intelligences
Different domains have their own types of content (e.g. visuospatial or verbal) and processes that structure and transform that content – what may be called domain-specific ‘syntax’. There is a syntax for how sentences may be constructed in meaningful ways. But there is also a kind of ‘syntax’ for how geometrical shapes may be visualized and transformed in the mind’s eye, and a ‘syntax’ for operations on numbers and symbols in mathematics, and a ‘syntax’ for how emotions work and so on. Individuals have different strengths and weaknesses in using content and ‘syntax’’ in these different domains to solve problems and address cognitive challenges, and this underpins the idea of multiple intelligences on this account.
Gf–Gc Knowledge Transfer
And there is also a transference process from Gf and Gc.
Much of our knowledge and skills stored in long-term memory that we use in problem solving and skillful action from day to day (Gc) was first derived by fluid intelligence (Gf). Cattell‘s ‘investment theory’(1971, 1987) states that individuals can use their fluid intelligence to apply to, or invest in, learning specific crystallized skills or knowledge. For example, when you grasp a new concept, rule or relation while solving a problem, or figure out a complex sequence or technique in a sensorimotor skill. these ‘outputs’ of fluid intelligence may be encoded in long-term memory neural circuitry, becoming hard-wired or ‘crystallized’.
Cognitive Load and Fluid Intelligence
Cognitive load is a measure of how demanding and effortful a task is executive processes – on the limited capacity bottleneck of working memory and attention. Tasks requiring fluid intelligence have higher cognitive load.
When a cognitive challenge that is the target of executive processes is novel and complex (e.g. matrices) then it has more cognitive load and requires more executive processing and fluid intelligence.
When the cognitive challenge that is the target of executive functions is already known and simpler (e.g. vocabulary or simple addition tests) then it has less cognitive load and requires less executive processing and fluid intelligence.
Controlled vs Automatic Processes
The executive processes that are more intensively tapped by fluid intelligence depend on brain networks involving the frontal lobes that are often called cognitive control networks.. Cognition can be classified as either controlled or automatic.
A controlled process is attention and working memory demanding. Controlled processes imply some degree of cognitive load. Controlled processes are under the flexible, intentional control of the individual, that he or she is consciously aware of, and that are effortful and constrained by the amount of working memory capacity.
An automatic process, on the other hand, can occur without the need for attention or limited working memory capacity. Automatic processing can occur without increasing cognitive load or interfering with other concurrent thought processes. When skills become well-practiced they may be automated. Automated processes such as well-learned skills that can be used to solve problems or attain goals are part of the store of crystallized intelligence (Gc).
For example, when you were learning to drive you had a high workload task to cope with. You needed to maintain and process critical information in working memory, paying careful, conscious attention attention to steering, using the mirrors, using the clutch & gears, accelerator and brakes. You also needed to continually update the contents of your working memory, monitoring where you are, the locations of other cars, traffic lights, road signs, pedestrians, and so on. Having a conversation with someone in the passenger seat was difficult, because talking requires controlled processing, and competes for the limited capacity (the bottleneck) of your executive processes.
But now, as an experienced driver, you can perform the basic skills of driving safely and navigating well-known routes both automatically. Much of your driving now occurs without your conscious awareness, and you can ‘dual task’ while driving – such as holding a conversation. But when there are new, unpracticed demands on your driving – e.g. navigating to somewhere new using a satnav – your driving requires more cognitive control, attention and working memory resources, once again competing with other tasks such as holding a conversation.
2 Comments
All comments welcome!
Great article.
Comments are closed.