A discussion of the role that context plays in habitual responding (smoking, eating & online shopping) and how this relates to both dual-system and rate-correlation theories

Introduction

Here we will consider the role of context and the evidence supporting its role in habits such as eating, smoking and online shopping. Most research has focused on eating behaviour; however, we are interested in the findings from eating, smoking and online shopping to understand if the mechanism is indeed standardised for three topographies. This is an important area for research given that we need to further understand the mechanisms of addiction and the relationship between different types of addiction, with particular focus on propensity towards addiction per se. It is proposed that the topography of habit is not as important as the process so, theoretically speaking, the process and findings should be similar across these types of habit. From the evidence, we will go on to consider the theoretical basis for explaining how a habit persists despite outcome devaluation, with special consideration of dual system theories and current thoughts on the mechanism of habitual responding. 

Goal: 

  1. Here we are planning the first study of its kind that explores the relationship between substance and non-substance-based addictions
  2. We are also looking at the role of primary (food) reinforcers versus secondary reinforcers (smoking, online shopping)
  3. We are looking at the role of single reinforcers (smoking) versus generalised reinforcers (online shopping) 
  4. We are looking at the difference between regular behaviours (food, online shopping) and more repetitive substance-based addictions
  5. We are also looking at primary reinforcer (food) versus a more generalised reinforcer (online shopping). 
  6. We are exploring the theoretical literature on habitual responding relating to both dual system and rate correlation responding, along with the evidence which supports this. 

Food and context

The majority of research has looked at the role of context in habits using food cues (Watson, Wiers, Hommel & de Wit, 2014). This is particularly salient given that there is a growing epidemic of obesity and across 34 countries, 18% of the population is now obese (OECD, 2013). Because obesity has knock on health consequences, it is even more important to understand the role of the environment around food motivated behaviour. 

Associative learning between cue and action is conditioned (Bouton, 2011) so, for example TV commercials will increase the consumption of the food they advertise independently of explicit intentions to diet. Pavlovian processes interfere with goal directed action and are insensitive to motivation and satiation. Cues which are associated to a certain outcome trigger responding even when satiated and the motivation for the food is reduced. The mechanism of Pavlovian Instrumental Transfer (PIT) contributes to excessive food motivated behaviour and explains how habit is transferred to for example McDonalds commercials. In situations where someone is reminded of burgers and the dual process model explains that freely elected behaviour is goal directed but the reward cues elicit reward seeking behaviours independent of the desirability of the outcome. There is, however, a reduction in responding which demonstrates an additive mechanism. The first response is expected to be insensitive to satiation but subsequent exposure in a sated state should allow us to adapt our behaviour but once we have successfully procured a certain food reward, we usually persist in consuming it. 

Thought 1 : 

  • Further research could examine the potentially differential effects of discrete versus more general contextual cues on the motivational modulation of instrumental behaviour. 
  • Would presenting the non-desirable food contingent upon responding gradually reduce the strength of the original S-O-R Stimulus Outcome Response) associations underlying the PIT effect. 

Measures used: 

  • Likert scales to rate hunger
  • Dutch Emotional Behaviour Questionnaire (DEBQ) van Strien, Frijters, Bergers, & Defares, 1986
  • Barrat Impulsivity Scale (BIS-11) high scores indicating impulsivity (Patton, Stanford, & Barratt, 1995) Stanford et al., 2009

These Pavlovian cues are persistent and when there is extended training, cue exposure or counter conditioning they cannot reduce the ability of the cue to elicit the response. This highlights the obstacles faced by people trying to lose weight and it is the associative mechanism which contributes to the high weight loss failure rate (Tsai & Wadden, 2005; Wing & Hill, 2001). An alternative approach is for people to form intention implementation plans. Often called “if-then” plans where they learn to identify the environmental cues to help plan healthy courses of action (Gollwitzer & Sheeran, 2006; Luszczynska, Sobczyk, & Abraham, 2007). This would lead to automatic performance of the desired behaviour when encountering these cues. The most effective way forward is to not learn them in the first place. Therefore, policy should focus on prevention, protection of children from unhealthy food adverts and regulation around food advertising, especially aimed at children (Hawkes & Lobstein, 2011; Matthews, 2008)

Thought 2: 

  • Investigate the efficacy of if-then plans in the context of PIT interactions given that research shows that once these associations between cues, responses and food outcomes have been formed, it is challenging to override or extinguish them (Bouton, 2011; Delamater, 1996; Hogart & Chase, 2011; Todd et al., 2012.  

Smoking and Context 

Related studies by Hogarth and colleagues (Hogarth, 2012; Hogarth & Chase, 2011) show that pictures of cigarettes and instrumental choices for rewards bias responding towards the picture outcome. Responding was not reduced by health warning exposure or dose of nicotine. The cues actually prime smoking behaviour and craving for cigarettes (Hitsman et al., 2013; Hogarth et al., 2010). Similarly, this was found independently of satiety of smoking or administration of a nicotine agonist (Hitsman et al., 2013). This transfer effect is also thought to play a role in food-seeking behaviours (Hogarth, 2012; Hogarth & Chase, 2011). The PIT effect in smokers shows that cue-inducted cigarette-seeking is not sensitive to health warnings of pharmacological intervention (Hitsman et al., 2013; Hogarth, 2012). Abnormal perseveration and delayed extinction in self administration is a sign of dependence (Deroche-Gamonet, Belin, & Piazza, 2004). Not being able to modify behaviour following a change in the contingency, abnormal learning and persevering despite adverse consequences is the hallmark of clinical dependence (APA, 1994). Behaviourally, dependence is defined by a greater control of drug seeking by Pavlovian paired stimuli (Di Chiara et al., 1999). There is a great cue effect observed with self-administration, natural rewards, attentional bias (EEG, fMRI) and over responding to cues related to instrumental drug seeking behaviour. In addition, a higher sensitivity to reinforcement, higher scores on subjective liking scales and more willing to pay/work drives up the rate of instrumental behaviour around drug seeking. 

Nicotine dependence is linked to greater tobacco preference supporting a value-based instrumental choice theory of dependence (Hursh & Silberberg, 2008). In addition, dependence is marked by a sensitivity to stimulus control on the drug seeking behaviour and whether specific PIT effects vary as a function of dependency status (e.g., Hogarth, Dickinson, Wright, Kouvaraki, & Duka, 2007). 

Drug-use history questionnaires used: 

  • Diagnostic and Statistical Manual of Mental Disorders nicotine dependence criteria (APA, 1994; Donny & Dierker, 2007) 
    • Cigarette Dependence Scale (CDS-5; Etter, Le Houezec, & Perneger, 2003); 
    • The brief Questionnaire of Smoking Urges (QSU; Cox, Tiffany, & Christen, 2001), with Factor 1 scores reflecting desire for positive tobacco reward and Factor 2 scores reflecting desire to avoid negative abstinence states, using the updated scoring system (Cappelleri et al., 2007); 
    • The Assessment of Substance Misuse in Adolescence (ASMA) scale, which uses a 0–9 scale and estimates illicit drug use and dependence (Willner, 2000)
    • The Alcohol Use Questionnaire, which estimates units of alcohol consumed per week and yields binge-drinking scores (Townshend & Duka, 2005). 

When there is a change in contingency, perseveration occurs in animals more vulnerable to dependence (Belin et al., 2008; Brenhouse & Andersen, 2008; Deroche-Gamonet et al., 2004; Diergaarde et al., 2008; Economidou et al., 2009), it is also a marker for exposure rather than dependence (although non-smoking control not compared). It is also possible that abstract rewards such as right or wrong in discrimination-reversal mean that perseveration is more easily detected, also in paradigms where self-administration of reinforcer occurs. The use of points might discourage perseveration. Is neurocognitive damage from drug exposure a factor in perseverative behaviour? However, the research does not support that there is a vulnerability for perseverance in people with nicotine dependence in relation to tobacco-seeking. 

Thought 3: 

  • Does this mechanism of perseverance still apply with context specific reinforcers whether self-administered or not?
  • Is the addition of an external reinforcer a factor in this?
  • Did this study replicate the true situation with regards to nicotine addiction?

Online shopping and context

There is limited research on the role of context in online shopping. Prashar, Vijay & Parsad (2017) examined the effects of online shopping values with website cues and their effect on purchasing behaviour using the S-O-R framework. There are likely differences in socio-cultural and demographic variables which influence buying and the current study was the first to look at an emerging market (India). The various antecedent factors, their significance and influence of online purchasing intentions were examined applying the concept of S-O-R. 

The current research looks at how people rate their internal and external factors, namely online shopping values (attitudes) and web environmental cues (context). This study contributes to the factors influencing shopping attitudes and the prominence of these factors. 

Through the S–O–R model, an attempt has been made to describe the impact of a store’s offline and online environment on consumers’ shopping behaviour. With stimuli, mediating variables, and behavioural outcomes, the S-O-R framework observes that the cues provided by a website effect attitude and response behaviour (Richard, 2005). In addition, Koo and Ju (2010) show that online cues impact emotions and intentions and Wang, Hernandex and Minor, (2010) find a connection between web aesthetics, quality of product and buying experience. Prashar, Vijay & Parsad (2017) examine the relationship between hedonic (related to S-R responding) and Utilitarian (related to goal directed responding) and the role cues on the intention to buy and stimuli with purchasing response behaviour.

The hedonic shopping values (HSV) and utilitarian shopping values (USV) are represented in the literature for retailing (Babin, Darden, & Griffin, 1994; Liu & Forsythe 2010; To, Liao, & Lin, 2007). The emotional and entertainment values have a hedonic tendency with enlarged arousal involvement with freedom, fantasy, emotive facets and escapism relating the buying experiences (Babin et al., 1994; Sorce, Perotti, & Widrick, 2005). Whereas the USV is related to shopping which reflects specific needs, reflecting a more goal-directed outcome (Babin et al., 1994). These two intrinsic and extrinsic values have been represented in the literature where studies show that buyers might have a goal directed buying motivation for hedonic products which have a positive influence on website satisfaction (WS), where HSV has a positive impact on WS and USV also has a positive influence on WS (Childers, Carr, Peck, and Carson, 2002; Chiou and Ting, 2011). It is found that perception, attitudes and what type of information provided and how it is presented are equally significant (Richard, 2005). The effectiveness of information (EIC) is used to determine what is pertinent for shoppers. Where WS has an intervening role on the relationship between online shopping values, and website atmospheric cues (inputs) and the intention to buy products (response) Bai et al., (2008). Website Entertainment (WE) has the strongest predictor of website satisfaction and this is closely followed by EIC. None of these predictors have a direct influence on purchase intention. Entertainment quotient is the most influencing element and the informativeness is least significant among cues. The entertainment value is more related to utilitarian rather than hedonic value (Jones, Reynolds, and Arnold (2006). Whereas, Stoel, Wickliffe, and Lee (2004) noted the opposite.  The utilitarian values are less than the hedonic values but are still significant in the Indian online market, highlighting the need to provide higher quality information. 

Questionnaire used:

Hedonistic & Utilitarian shopping scales (Babin et al., 1994) 

Thought 4: 

The present study had not taken into account the issue of shopping context, which shoppers might consider. Future studies can incorporate different shopping contexts and examine whether there is any change in consumers’ response. 

Thought 5: 

Does training using response outcome over a period shift the balance towards more context driven responding on a simple laboratory task. Do utilitarian and hedonic products influence this when considering buying for example “washing up liquid” versus “a pair of designer shoes”. 

Thought 6:

They did not use context cues, the SOR was not the correct one used in psychology and so far, no research actually focuses on this behaviour. 

Mechanisms of Addiction

The addiction literature is dominated by the dual process additive model. This is proposed due to the fact that both goal-directed and habitual controller operate in parallel to each other in relation to the final course of action. There appears to be an additive response to a cue test, which supports this theory and it is thought that there is an evolutionary adaptive process which leads to automatic responding. This is because cues signal food availability and the mechanism is automatic and linked to our need for survival. Because of the abundance of food, the environment being saturated with food-associated cues, this process has somewhat turned against us. These Pavlovian instrumental interactions which are insensitive to satiety could contribute to our excessive food-seeking and the recent rise in obesity and addiction to other substances. However, these cues also interfere with our immediate and flexible adaptation to the change in desirability to food rewards (Holland, 2004; Rescorla, 1994). 

Thought 7:

Would presenting the no-longer desirable food contingent on responding reduce the strength of the S-O-R associations and the underlying PIT effect?

Thought 8: 

Would presenting the desirable substance non-contingently weaken the effect of certain cues to responding?

Thought 9: 

Would presenting the response still occur if the outcome was no longer available. Where does this behaviour go? Would making a new behaviour contingent on another cue mean that the old behaviour no longer exists, or would it still cue the response? 

Cue-induced processes initially interfere with learning processes and it is expected that the first response is not sensitive to satiation, however, subsequent ones should be able to allow us to adapt our behaviour. It is thought that this is too late by then, as the seeking, procurement behaviour then leads to consumption anyway, so we usually persist in consuming it. This highlights the difficulty of people trying to lose weight and these associative mechanisms contribute to the failure of people when trying to lose weight (Tsai & Wadden, 2005; Wing & Hill, 2001). Similarly, this is also found in people who smoke, and investigations of cue-induced cigarette seeking is not sensitive to health warnings (Hogarth & Chase, 2011) or pharmacotherapy (Hitsman et al., 2013; Hogarth, 2012). 

The dual system proposes that both the goal-directed and habitual controllers run in parallel to one another in an additive manner which determines the final course of action. The freely elected behaviour is goal-directed and the cues in the environment which are associated with it are elicited and this is independent of the desirability of the outcome. The instrumental performance is related to the strength of the response outcome (R-O) and the value of the response in reinforcement learning theory (Mackintosh & Dickinson, 1979). Interestingly, studies show that the use of free operant schedules with the motivation re-established tell us that reward probability is not likely to be the primary determinant in goal-directed control. 

Reinforcement Schedules

Variable Interval (VI) schedules are used in habit research as these are found to be the most appropriate for habit formation. This is thought to be because they model a resource. For example, nectar is something which depletes when taken and over time it regenerates. The VI schedule is where the outcome is available on a time dependent manner and VI-10s is used in current research (Tricomi et al., 2009) and an average interval has to elapse before the next one is available. A Variable Ratio (VR) schedule is similar to the behaviour you might see in a foraging animal where resources do not deplete, and each action has a fixed probability of gaining access to the resource which is independent of the time elapsed.

An interesting experiment carried out by Dickinson et al., (1983) which matched the outcome probability of interval and ratio schedules found that once the outcome had been devaluated the ratio trained rats decrease in performance but this did not happen in the interval trained, even though instrumental responses were lower in this group. These findings have been extensively replicated in other studies (Gremel & Costa, 2013; Hilario et al., 2012; Wiltgen et al., 2012).  In fictional investment tasks, ratio training gives higher causal judgments on key pressing tasks (Reed, 2001). However, interval contingencies are found to differential reinforce the time between the pauses and the probability of reinforcement increases with the time between responses. This is a monotonic trend and faster responses occur with shorter programmed intervals between rewards. Ratio responding is not dependent on the time that elapses between pauses so there is a fixed probability of reward. The reduced sensitivity to devaluation of the outcome with two interval choices is not explained when there are two outcome choices with two interval sources, because the performance is highly sensitive to outcome devaluation (Kosaki & Dickinson, 2010). It was found that when there were two levers and two choices, devaluing one of the outcomes decreased performance of the corresponding response, even when there was only a single lever. This contrasts with insensitivity to outcome devaluation when matched training occurs with a single response. 

It is unclear why lever pressing is not affected by contingent or non-contingent outcomes once it has been devaluated and why choice and single response training affects the goal directed responding when it is being differentially reinforced.

Thought 10:

Lever pressing for instrumental responding for two outcomes is non-contingently delivered and not affected once the outcome has been devalued so the response-outcome feedback rates are the next distinction to be considered in interval and ratio schedules.

Dickinson (1985; Dickinson & Perez, 2018) suggest that R-O learning is driven by the rate of correlation and the experience of the responder. The greater the experience rate, the stronger the R-O outcome in the correlational law of effect. This is a linear correlation with the outcome rates established by the schedules. Outcome devaluation for goal-directed responses have been confirmed (Urcelay & Joknman, 2019) who found that delaying the food outcome by a 20s interval meant that it is no longer sensitive to outcome devaluation compared to those with no delay. 

Thought 11: 

Is this experiment not just uncoupling of the response outcome associated context cues? Does this then shift the responding to more cue-response, habit related responding?

Contingency degradation

If you vary the probability of contiguous outcomes and the likelihood it will occur in the absence of non-contiguous outcome, then there is no control over the number of outcomes which are received in a given time period. If these are equal these do not sustain instrumental responding that is established without contiguity to the outcome. There is contingency degradation (Shanks, 1991) and the judgement of the R-O outcome association and rate of responding is lower when this is degraded by the probability of NCO (non-contingent outcome). This explains how the ratio and interval schedules are in effect when the rate correlation probabilities are matched and the degradation of the causal contingency between the R-O happens and is the theory for any goal-directed control to occur. 

An additional learning system is proposed to give us a complete account of instrumental behaviour because we are not left with a complete account of sustained responding on an interval schedule. Because there is a low correlation experienced under this schedule it does not explain why responding extinguishes when outcomes are withheld. A previous learning experience is required to associate and recycle. The outcomes are therefore present in memory (which is also the case during extinction), therefore an additional learning system is required. 

Dual system theories

Reinforcement Learning (RL) systems aim to maximise the number of rewards obtained during a task (Daw et al., 2005; Dolan & Dayan, 2013; Keramati et al., 2011). Where Model-based is the same as the goal-directed system and Model-free is not sensitive to outcome revaluation and looks at the running average of reinforcement from each action from action/state values which is faster and less expensive than the Model-based system. In the Model-free system the action is contingent on the consequences, is an important aspect of behaviour control and similar to habit responding (Niv et al., 2005; 2007). This habitual system makes a distinction between ratio and interval contingencies using an economic argument between the utility by responding and the cost of emitting the response, leading to a strategy which obtains the most outcomes with the least effort. However, this Model-free system does not account for the differential sensitivity of ratio and interval responding in outcome revaluation experiments. Evidence for the dual system looking at interval and ratio responding shows that when half of each of these groups were devalued, the ratio responding decreased, indicating goal-directed responding.  The interval responding was unaffected by devaluation but the level of ratio responding did not differ from the interval group because the outcome probability was matched. 

Thought 12:

The habit system should have been equal in both groups because when responding is sensitive to devaluation it is goal directed. However, this could not be attributed to devaluation in ratio group and this is not attributed to the bait component and implies both systems sum the strength to determine the probability of responses. Therefore, when interpreting divergent results between ratio or interval it is not the amount of training which determines responding to become autonomous of the outcome value but whether the mechanism of memory recycling yields a lower local rate correlation. 

Choice responding

Habits are strengthened in the case of interval schedules by the temporal control of outcome availability in free operant or concurrent schedules. The memory cycles will calculate that some samples will have no response or outcomes of a switched source and others will contain these representations. The switched source as a consequence leads to sustained rate correlations for both of the responses which will sustain goal directed control. If one wants to reduce goal-directed strength, then a switch from a contingent to a non-contingent schedule would lead to responding insensitive to devaluation and the results of this have not yet been reported. As habit learning is explained as a form of S-R learning then we are expecting there will be sensitivity to temporal cues registered since last response and this impacts rate of responding. In the ratio schedule the probability of reinforcement is independent of the IRT and the size of the emitted IRT. Kuch & Platt (1976) specified a schedule referred to as the regulated-probability interval schedule (RPI). This is a unique kind of schedule which sets the probability of reinforcement for the next response so that if the behaver continues responding at the current rate the outcome will match that specified by the scheduled interval parameter. Therefore, variations in the rate of responding will not have much impact on the outcome rate. The dual system says there will be no difference in sustained responding on the ratio and RPI. Tanno & Sakagami (2008), did not observe a difference between RPI and ratio responding but did for standard matched intervals (RI & RPI where RPI had higher response rate), where they had a reduced responding. There are also reports that ratio responding higher than yoked RPI (Perez et al., 2018) which conflicts with Tanno and Sakagami’s (2008) findings.  

Discriminative control

The dual system model does not explain how the strength of the goal is a product of rate correlation between the responses and outcomes. Colwill & Rescorla (1991) trained rats to emit two different responses and outcomes, the stimuli to signal the outcome is produced by each response. It was found that when one of the outcomes was devalued, the rates were higher in extinction than when not devalued (during training) giving evidence of the S-O associations on the stimuli and R-O association on the responses, suggesting a relationship between the SRO. 

Thought 13:

Why is this not incorporated into the rate correlation theory?

After instrumental training, it was found that the devaluation effect was not affected by the shift from training to an unfamiliar context between the training and testing conditions (Thrailkill & Bouton, 2015). The rats were not more likely to discriminate between the context with extended training, especially as responding was autonomous of the outcome value and this represented how the context shift reduced overall responding. 

Thought 14:

Study 2 looking at the role of context manipulation with limited and extended training expects us to see that with limited training the responding is under goal-directed control for the R-O relations. This control transfers spontaneously across contexts as anticipated by the rate-correlation theory.

Thought 15:

When the response is under habitual control with further extended training the responses decrease because the control reflects the context development of S-R strength.

Motivating habits

The transfer two-process theory states that Pavlovian conditioning is to contextual cues and it occurs with instrumental learning in operation training. This implies that context and motivation therefore influences performance. Dickinson & Dawson (1987) trained hungry rats to press a lever for food and sugar water with a different context. Then the rats were given the lever when thirsty, despite no instrumental response had been trained in this condition and the lever was associated and trained with the pellet outcome in a hungry state. They found that the response trained with the reinforcer has a motivational influence even when the reinforcer is different from trained cues. There is thought to be a habit response with goal-directedness. However, training in hunger and testing in thirst shows that outcome revaluation does not indicate goal-directed control because it did not represent an action-outcome contingency because the lever press was trained with food pellets and not sugar water. This means this result shows that sensitivity of Pavlovian motivation to outcome revaluation leads to an erroneous attribution of goal-directed status (Corbit et al., 2007). 

Thought 16:

Was the context the same for both? What influence does the specificity of context have on instrumental responding? Does responding in the absence of motivation affect response when both are devalued?

Incentive learning

Dickinson & Dawson, (1988;1989) tested how incentives contrast the Pavlovian motivational control of habits which suggest that this type of learning is needed to function as goals for instrumental action. Habits are learned when the incentive value conditions when for example drinking sugar water when thirsty. It is these hedonic reactions that cause us to learn through evaluation of the outcome. This means that habits are motivated through conditioned drive towards a contextually eliciting stimuli and referred to as motivated goal-directed action.  

Thought 17:

Is this not sometimes called attentional habit? People need to learn about the incentive values of these outcomes when in different motivational states before they can gain goal-directed action. 

Evidence for feeding coming under the control of contextual stimuli comes from a study carried out by Capaldi & Myers (1978) where rats ate pellets when they had extensive training under hunger, even though they had been sated on the pellets in their home cages. 

Thought 18:

If a value incentive is placed onto a particular food, we would expect a strong learning effect when there is exposure to the pellets in the low (un-deprived state) in the operant chamber rather than the feeding cage. Why is the incentive different in the operant chamber compared to feeding cage? 

To determine this effect, half of the rats were re-exposed in the operant chamber (rather than feeding chambers) to test if animals in the un-deprived state press the lever less than those re-exposed when they were hungry. It was found that un-deprived animals press just as frequently as the hungry ones when there is no prior experience of food outcome in the un-deprived state. Therefore, for the motivational state to have control over the instrumental performance there has to be prior experience of the outcome in this state and is in line with the predictions of incentive learning theory. 

Summary

The rate correlational theory assimilated into the dual-system framework gives us an account of both goal-directed and habitual control. The dual system theory which incorporates the rate correlation system does not account for the way that goal-directed and habitual system work in summation to generate behaviour in psychological or behavioural terminology. It is possible that the account of both these systems working in synergy can be explained by considering an instigation (goal-directed) process and then an execution (which could be both goal-directed and habitual). This could be a cascade of associative interconnections between both systems in synergy and it is suggested that this is best approached with the use of a cybernetic model (Dickinson, 2012). The instigation could represent the latent intention, which retrieves the belief about the causal consequences of action between many response outcome representations (Dickinson, 2012). The activation process has causal consequences of action due to an association between the response and outcomes. The association between the representation of the outcome, the motivational mechanism and the feedback process completes the practical inference to generate an executable intention.

The failure to detect a revaluation paradigm indicates habitual responding and the opposite goal-directed control. They have also been disassociated motivationally with habit being more sensitive to motivational effects of the context through Pavlovian condition. Because habits in this form of motivational effect has a sense of goal directedness because they are sensitive to revaluation and deepened by the sensory representation of the outcome which also acts as a stimulus which elicits the response through the S-O-R chain generated by S-R learning. The motivational and stimulus effects do not fulfil the criterion because they do not operate through the causal representation of the R-O contingency. The interaction between the goal-directed and habitual control is not accounted for in the model-free and model-based system or explain how interval contingencies establish habitual control. It also suggests goal-directed control occurs through rate-correlation. 

The foregoing are the conclusions of the author of this paper. Comments and feedback are invited to: info@simplybehaviour.com

Want to learn more about applied behaviour analysis?

Why not sign up to our graduate level applied behaviour analysis practitioner (level 1) course. This is an online, exam assessed course covering the core principles of applied behaviour analysis. It is accredited by the AABP and meets the core sequence requirements for certification as a behaviour professional. You will be an ABAP-1 on successful completion of this course sequence and will gain your certificate and your accredited logo: 

Further Courses, all of our courses are certificated and gain you CEU's
Some Of Our Other Articles

To sign up to our newsletter through Facebook click here: 

Simply Behaviours Facebook Page

svsporngames.com