Behaviour Change

There are many flavours of behaviour change – some rely on persuasion and changing attitudes, others on goal-setting and implementation intentions. The behaviour change that I am interested in works at a more fundamental level and enables change by communicating directly with systems that control behaviour. In many ways, behaviour change can work best when one bypasses the cognitive system and work directly at developing new healthy and positive behaviours. In this way, issues of reactance and attitude can be side-stepped.
The principles of behaviour change (BC) are based on many years of both behavioural and neuroscientific research. Indeed the techniques that BC has adopted have been taken from fundamental research into how animals (including humans) learn and make decisions, and develop and maintain behaviour. Essentially, neuroscience research has confirmed the power of BC techniques as well as developing our understanding of how BC works and why other techniques to change behaviour have not been so successful.
Policy-makers have realized that just because an individual has the intention to change, this may not actually result in adaptations to their behaviour. This Intention-Action gap is a result of several factors inherent to human behaviour. For example, consider attitudes towards energy awareness and sustainability – most people adopt ‘sustainable intentions’ such as switching off lights, recycling bottles etc. But far fewer actually develop and maintain these new behaviours. One reason for this is that even though individuals maintain an explicit intention to behave in a new or different way, they experience a difficulty in suppressing old habits and so find themselves leaving lights on and throwing bottles in the waste bin. As expanded below, neuroscience research has demonstrated the existence of separable neural systems that control habitual versus intentional behaviour, thus giving insight into why we exhibit intention-action gaps, as well as providing direction in how to close the gap and behave more as we would intend. In our modern world with an increasing prevalence of obesity and also of sedentary behaviour, closing the intention-action gap in the health domain is more important then ever.
Dual-process model of behaviour
Squire et al., (1990) famously proposed the distinction between Declarative and Procedural memory. The declarative system represents explicit memories of events and meanings that could be consciously recalled and manipulated. In contrast, the procedural system represents skills and habits that are implicit in nature (the representations are not accessible to conscious awareness). Since this initial distinction as to the way in which we represent the world, the idea that two analogous systems control behaviour has developed. Broadly speaking, most models postulate two levels, each processing different types of information for different aspects of behavioural control. One system is intentional, effortful and propositional; the other is implicit, routine and uses affective signals as its currency (Evans, 2008). From an adaptive perspective it has been argued that two systems evolved in order that routine behaviour could be maintained whilst allowing the parallel engagement of cognitive processing. So, the implicit system gradually acquires an enormous data-set of routine operations allowing the cognitive system to be free, flexible and accessible. Research into human motivation concurs with this conceptual framework: it has been shown that an implicit motive system controls our spontaneous everyday drives, whilst an explicit motive system regulates this by incorporating social norms and conscious expectations on our goal choices. It turns out that implicit motives are better predictors of future performance (McClelland, 1991). Likewise, implicit processes such as habits and routines control much of our daily behaviour without a great deal of conscious or explicit input. This is fine when habits are adaptive, but challenging when they produce inappropriate behaviour.

The neuroscience of dual-process behavioural control

In the early 1900s, Thorndike developed his Law of Effect (Hernstein, 1970) describing how behaviours that resulted in a positive consequence tended to be repeated, whilst those that didn’t (or were punished) were not repeated. Interestingly, the underlying representations were argued to be Stimulus -> Response based (S-R) such that environmental triggers (the S) reflexively produced the motor output (the R). Research supports this by showing that as a particular behaviour becomes over-trained it shows less flexibility is simply transformed into a stimulus-response map (Dickinson 1985; Adams and Dickinson 1981). These findings, resemble routine (or procedural) learning closely and have recently led to neuroscience work supporting the existence of S-R maps in brain systems. For example, fMRI work has shown that the training of a habitual response leads to a gradual increase in brain activity in the putamen region of the stiatum (as responding becomes more under the control of the triggering stimulus; Tricomi et al., 2009). Behavioural neuroscience work has shown that damage to the striatum in rats also impairs the development of habits and that the cortical afferents to the striatum also support habitual processing (Balleine and O’Doherty, 2010).
Alongside this, work has also demonstrated the existence of a parallel explicit goal-directed action system, and more importantly, that interactions between these explicit and implicit systems underlie behavioural choice. For example, prior to over-training, behaviour is tightly linked to the value of the goal and so fluctuations in the goal value change behaviour (Dickinson 1985) – in other words, if are hungry and want chips, you’ll but chips. Following over-training, changes in goal value do not affect performance – the behaviour has become inflexible and automatically triggered by the stimulus – even if you are not hungry, the sight of chips will induce you to buy them. Further, research in patients has provided evidence of double dissociations between the two systems (Knowlton et al., 1996). Parkinson’s patients have a dysfunctional striatum but intact temporal lobe, whereas amnesics often have damage to the temporal lobe, but their striatum is intact. The striatum is implicated in habitual behaviours whereas the temporal lobe supports declarative memory. Knowlton and colleagues tested both amnesics and Parkinson’s patients on a probabilistic classification task which involved both explicit and implicit aspects. Amnesic patients could not recall the training and yet were able to perform the task successfully (explicit impaired, implicit intact). In contrast, Parkinson’s patients could recall the training but could not learn to complete the task (explicit intact, implicit impaired). This double dissociation supports the existence of separable explicit and implicit learning systems subserved by distinct neural circuitry. Research has also demonstrated that communication between the two systems may result from prefrontal projections onto the striatum – certain regions of the prefrontal cortex have been argued to provide inhibitory control (or will power) over pre-potent automatic responses (Aron et al., 2003).
In summary, for complex behavioural routines, it seems that distinct cortical and subcortical networks underlie the way in which explicit goal-directed actions and implicit habits compete for behavioural control (Parkinson et al. 2000, Balleine and O’Doherty 2010). Our developing understanding of these circuits provides insight into the nature of psychological deficits in patient groups but also enables a better understanding of how behaviour is learned, maintained and changed – for example, the amnesics in the Knowlton study had no explicit awareness of learning about task parameters and yet were as able as controls to successfully complete the task. Sometimes, learning proceeds without our conscious awareness and so behaviour change techniques should heed this when targeting intervention.


As Thorndike famously argued, when our actions have positive consequences we are more likely to repeat them. No matter how convoluted our cognitive rationalization for behaviour, we should accept the truth that a lot of what we do in our everyday lives, we do so because we have been previously rewarded. There is a long history of animal learning theory and behavioural neuroscience research studying reward and incentive-driven behaviour (Berridge, 2004). Broadly speaking, rewards promote learning and the more salient the reward the faster the learning. Further, rewards can be primary (food, sex etc) or secondary (more abstract incentives such as ‘art’ or ‘health’) and both can support highly motivated behaviour (Cardinal et al., 2002). This literature has been mirrored more recently with neuroimaging work looking at the neural basis of incentive learning and rewards in humans. For example, the anticipation and receipt of rewards are represented in a distributed network (Arana et al., 2003, Bray et al., 2010) both cortically and subcortically, perhaps reflecting the cognitive and semantic aspects of rewards and the hedonic and affective components respectively. More interestingly, there is evidence of a distinction between implicit and explicit representations of rewards. Johnsrude et al., (2000) asked participants to complete a working memory task using simple patterned stimuli. Correct responses to certain stimuli (though not others) were rewarded with chocolate. Following testing participants rated the patterns for aesthetic preference and gave reasons for those preferences. Participants preferred the chocolate-reinforced pattern significantly more than control patterns but were not aware of the reward association, instead they gave self-rationalized aesthetic explanations for their preferences. In other words, individuals were not aware that their subjective preferences had been shaped by simple food reward. In the same study, individuals with damage to their amygdala showed intact working memory performance but impaired preferences, whilst individuals with prefrontal damage showed impaired working memory but intact preferences. These data were corroborated subsequently in an fMRI experiment with healthy controls (Cox et al., 2005).
Using representations of reward to guide actions depends upon anticipation and expectation. Much work has focused on how affective anticipation guides behaviour and demonstrates that a powerful implicit brain network provides rapid heuristic calculations of value in order to guide behaviour (Knutson 2011, Damasio 1994). Critically, this system operates below consciousness (is implicit) and is based on the previous reward history of actions and goals. For example, using the Iowa Gambling task, Damasio and colleagues have demonstrated that individuals use ‘gut feelings’ (physiological signals of anticipated reward) in order to guide decisions in conditions of uncertainty (Bechara, et al., 2005). They have also demonstrated that decision behaviour becomes adaptive (participants begin to learn the task and act successfully) prior to becoming consciously aware of task contingencies. Damage to the OFC or amygdala impairs learning and performance on this task (Bechara et al., 1999).
In summary, a network of brain structures including the OFC, amygdala and striatum appear to represent the value of rewards and to feed this information directly to shaping adaptive behaviour. Clearly, humans can also explicitly value and state their preferences for goals and use this to inform behaviour; however, what the above neuroscience research demonstrates is that a powerful implicit reward system exists which can learn and shape behaviour without the conscious awareness of the individual; this system likely contributes to the development of routines and habits as described above.
Ever since the seminal work of Bandura (1977) in demonstrating the power of role-models in shaping the behaviour of children, much work has focused on understanding the processes underlying this form of learning and imitation. Recent neuroscientific study has provided significant insight into the nature of this process. Rizzolatti and colleagues (Rizzolatti et al., 2004) famously discovered ‘mirror neurons’ in the brain of monkeys – neurons which not only activate when a monkeys is making a particular behavioural response, but also fire when the monkey is watching another make those same actions. These neurons, which have been observed in human brain networks using fMRI, appear to be part of an action-observation and imitation system which has been argued to underlie the way in which we understand the actions of others and also to aid us in learning to perform those actions. Such a system is likely the basis for higher-order social processes such as empathy and role-modelling. Subsequent work has shown that we are also more fluent at performing actions when we have watched an actor perform the same, rather than an incompatible, action (Jackson et al., 2006). This work has demonstrated that imitation is both implicit (we are not necessarily consciously aware of the fluency) and pervasive (we imitate others’ actions without any explicit intention to do so). This work, taken together with social neuroscience research on how specific role-models are identified, suggests that humans have an inherent tendency to adopt the behaviours of significant others and suggests that we develop behavioural habits both through experiencing direct rewards ourselves and also by observing others do so. Contemporary neuroscience research is currently focusing on the specific neural systems that underlie this mechanism and also how it might malfunction is certain clinical groups such as autism (Lacoboni, 2005). Imitation works in a similar manner to rewards in that the observation of others’ behaviour provides implicit social cues as to how to behave and as such shape our responses within a complex environment.
It can be seen that the human brain has evolved a set of implicit and semi-automatic processes that enable adaptive behaviour without the need for much cognitive or conscious input. Reward-learning allows us to accurately predict and navigate the world adaptively, whilst imitation and role-modelling provide social cues to a similar end. Both of these systems provide teaching signals to develop and maintain behavioural routines which are triggered by reward or contextual stimuli and ultimately guide us through a complex world in our attempts to achieve goals. Behaviour can be seen, not as a series of distinct steps, but as a dynamic stream of responses to changing stimuli in the environment which can operate outside of awareness and are robust in their power to control our daily lives. Behaviour change then is not so much a endeavour to simply alter peoples’ attitudes but instead to bypass the cognitive and speak directly to the systems that control much of our ‘spontaneous’ actions across the day. Recent neuroscience work has gone a long way to helping understand how the brain supports these systems and also gives insight into how we might develop interventions that help individuals in realizing healthy and long-lasting behavioural changes.


There are excellent articles developing some of these themes, so for some further reading:

How to design behaviour change interventions (using a combination of design thinking, complexity theory and nudge psychology):

Parkinson JA, Eccles K and Goodman A (2014) Positive impact by design: the wales centre for behaviour change. Journal of Positive Psychology 9 (6), 517-522.

More about stimulus control and the brain mechanisms involved:

Levant B., and Parkinson J.A. (2014) Positive emotions and reward: Appetitive systems – Amygdala and striatum. Reference Module in Biomedical Sciences. Elsevier. 24-Oct-14 doi: 10.1016/B978-0- 12-801238-3.04498-6.

Self-Determination Theory (the importance of autonomy, competence and relatedness):

Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. American Psychologist, 55, 68-78.

The intention-action gap, why actions are more important than just talking a good game:

Sheldon, K. M., & Krieger, L. (2014). Walking the talk: Value importance, value enactment, and well-being. Motivation and Emotion, 38, 609-619. doi: 10.1007/s11031-014-9424-3

Leave a Reply