Over the past 10 years there has been a growing body of scientific evidence that a specific type of computerized cognitive training (CCT) has the potential to raise IQ significantly – namely working memory training. Commercial CCT comes in many forms. Many brain training companies divide training between different types of cognitive processing such as speed, memory, attention, and so on. In the light of all the studies done on these different types of training, it is only working memory training that has demonstrated real potential for increasing intelligence.
What is working memory?
Working memory can be thought of as our ‘mental workspace’. Imagine your mind as a white board that is constantly being written on, while you organize and do calculations with the material before erasing it as you shift your focus from one task to the next – that is your working memory system.
More formally, working memory is a short term memory and management system that
“provides temporary storage and manipulation of the information necessary for such complex cognitive tasks as language comprehension, learning, and reasoning.” (Baddeley, 2003, p. 189)
Individuals differ in their working memory capacity – i.e. the quantity of information they can maintain and process. Working memory capacity correlates highly with IQ.
This makes sense, as you can imagine that more intelligent people have a larger mental ‘workspace’ and are able to make relatively more connections and inferences using this spacious workspace.
Dual N-Back Training
The working memory exercise Jaeggi used was the dual n-back – which is now the most widely studied computerized cognitive training game. It involves viewing a continuous stream of items such as moving squares and deciding whether each square location matches the location ‘n’ stimuli back in time. The memory ‘gap’ is the ‘n-back level’. If you need to keep track of the location 2 moves back, then you are at the 2-back level. If you are tracking the location 3 moves back – then you are at the 3-back level. Here you can see a 2-back example:
In dual n-back training, both a verbal and a visual stream of items are presented simultaneously as shown below for a 2-back game:
How Effective Is Dual N-Back Training?
Jaeggi and her colleagues published their dual n-back study back in 2008. Since then numerous studies investigating the effects of dual n-back training on cognitive performance have been published, and there has been a heated controversy among both cognitive scientists and the popular media surrounding the idea that cognitive training can increase IQ.
Some studies have failed to replicate the IQ-boosting training effect – such as this study by a team led by Todd Thompson in MIT’s Department of Brain and Cognitive Sciences which concluded:
“[our] findings fail to support the idea that adaptive working memory training in healthy young adults enhances working memory capacity in non-trained tasks, fluid intelligence, or other measures of cognitive abilities.” (2013)
But numerous other studies find a significant IQ increasing effect, such as this one by Sarah Rudebeck in Oxford University’s Department of Experimental Psychology:
“we found that the trainers, compared to non-trainers, exhibited a significant improvement in fluid intelligence after 20 days…. Our findings demonstrate that practice on a …working memory task can potentially improve aspects of both ….memory and fluid intelligence.” (2012)
So how are we to evaluate the conflicting evidence? Does dual n-back training work or not?
In answering this question, we need to distinguish between different grades of evidence.
- ‘In house’ research conference talks, papers ‘under review’ and other non-published material. (e.g. Lumos Labs unpublished papers.)
- Single peer-reviewed journal articles that have not been replicated.
- Several peer-reviewed journal articles from different labs with replications.
- Meta-reviews of multiple peer-reviewed journal articles with replications.
The higher the level the more reliable the evidence – whether for computerized cognitive training, or for other brain cross training methods such as nootropics, intermittent fasting or meditation.
The two latest meta-reviews of working memory training – one published this year and the other being written up for publication – both conclude dual n-back training is effective in improving general cognitive performance and IQ test scores.
The 2014 paper by Jacky Au and colleagues at the University of California, concludes:
“Our work demonstrates the efficacy of several weeks of n-back training in improving performance on measures of Gf [fluid intelligence]. We urge that future studies move beyond attempts to answer the simple question of whether or not there is transfer and, instead, seek to explore the nature and extent of how these improved test scores may reflect “true” improvements in Gf that can translate into practical, real-world settings.”
My grad school colleague Jason Chein – Principle Investigator at the Temple University Neurocognition Lab reached a similar conclusion in his 2014 meta-review of working memory training, concluding that training “does indeed show positive transfer, even to far transfer measures” (personal correspondence).
Based on this meta-review evidence – the highest grade scientific evidence we have – we can conclude that dual n-back training is worth investing your time and effort into if you want to achieve gains in general cognitive performance. No equivalent gains result from other types of computerized cognitive training – or learning musical instruments, playing video games, or doing regular aerobic exercise.
Increasing IQ Gains: Dual N-Back Optimization
Jacky Au and colleagues argue in their meta-review that the average increase in IQ from training they found is an under-estimate due to the samples and testing criteria. Moreover, they believe that the effect size they reported could be increased by optimizing certain game parameters including:
- Program completion
- Increasing intrinsic motivation for program completion
For these reasons the authors argue:
“the results reported in this meta-analysis represent a low-end estimate of the true extent of improvement that n-back training can have on measures of [intelligence]”.
The name of the game is now is determining precisely what parameters of dual n-back training can optimize wide transfer to IQ and general cognitive performance. An analogy is growing tomatoes. If you know you can grow them, your next concern is to figure out how to make them bigger and better – by e.g. different watering schedules or adding fertilizer or varying the growing temperature!
Interference Control & ‘Second Generation’ Dual N-Back Training
One strong candidate parameter for optimizing dual n-back training that I have been particularly interested in is interference control.
Interference is a technical term for distracting information that is similar to the information you need to perform well in a game or cognitive challenge. If you are playing the n-back game, and you are at a 3-back level, a matching stimulus for N=2 or N=4 would qualify as interference. This is shown in the diagram.
Another example of interference is found in the ‘Stroop task’. You can try the task for yourself now – going from left to right, as quickly as you can say aloud the ink color of the words here – ignoring the word meaning (the answer for the first one is ‘red’).
Because reading words is so well-practiced and automatic, there is a strong interference effect in this task – as you’ll have noticed! The meaning of the word captures your attention, and interferes with your ability to perform the task. It takes concentration to control this interference.
There is good scientific evidence that interference control – the ability to filter out distracting information of this sort – underlies the link between working memory and intelligence.
- First, brain imaging studies reveal that neural mechanisms of interference control underlie the relationship between fluid intelligence and working memory span.
- Second, Claudia von Bastiana and Klaus Oberauera at the Department of Psychology, University of Zurich, have found that a game based on ‘supervision’ working memory training alone resulted in IQ gains. Supervision is terminology for that aspect of working memory used for interference control – defined as “selective activation of relevant and inhibition of irrelevant information”.
Based on the strong potential of interference control for IQ gains it is now built into all the IQ Mindware apps as a default. Building interference control into the dual n-back game is something new, and for this reason I call it ‘second generation’ (2G) dual n-back training.
Feedback from users’ experience of cognitive gains from the Stroop Dual N-Back has so been positive. This dual n-back variation has a great potential for optimizing IQ gains. IQ Prime also offers standard dual n-back training with interference.
In summary, we can consider an analogy with sports performance.
Substantial gains in sports performance are now possible – and achieving those gains relies on scientifically guided training programs. Likewise it is becoming apparent that cognitive performance gains are possible and these too rely on scientifically guided training programs. Evolved versions of the dual n-back and other forms of working memory training will be central over the next decade in achieving and maintaining the highest levels of cognitive performance.
