BLOG

Share 0
Like 0
Animal models in gambling research: in search of the jackpot…

Animal models in gambling research: in search of the jackpot…

At a recent meeting in Nantes on behavioural addictions I gave a presentation on animal models in gambling research. This column reflects this lecture along with some after thoughts.

Animal models in gambling research: in search of the jackpot…

 

At a recent meeting in Nantes on behavioural addictions I gave a presentation on animal models in gambling research. This column reflects this lecture along with some after thoughts.

 

Disordered gambling is a major societal problem with high costs to the subjects involved. Animal models may help in understanding and preventing gambling-related problems. Data show that some animals may have a gambling-like profile, yet we need additional steps to come to models of disordered gambling.

 

While gambling remains a recreational activity for most people, it may become a salient problem for some, developing into disordered or pathological gambling, i.e. subjects continue to gamble despite the fact that this activity disrupts their personal, professional or financial life. The prevalence of pathological gambling will likely increase in years to come due to the increasing possibility of on-line gambling through the Internet and, in general, more liberal views on gambling as activity. Animal models may be in different ways of help in the field of gambling research, e.g. as means for testing the effects of potential new drugs on gambling-like behaviour, assessing traits associated with the likelihood of engaging into gambling or developing disordered gambling, and exploring the key features of developing disordered gambling-like behaviour. Here, I will highlight a few aspects of models used.

 

Gambling is hooking onto or exploiting the workings of brain circuits that are designed or have evolved in the earliest vertebrates to deal with rewards and punishments, reduction of uncertainty through interaction with the environment (learning and exploration) and maintaining long-term efficient behaviour (control), while attending to physiological needs. Critical structures herein are prefrontal-striatal areas with associated structures such as the amygdala and hippocampus, in humans more developed than in say rats. Critical neurotransmitters herein are dopamine and serotonin. In all studies on disordered gambling in humans, structures within these circuits pop-up. While it is clear that some elements of gambling are not easy to model in animals, at least not in rats, such as elaborate reflective thoughts on contingencies, others certainly can. In fact, illusions of control are at their basis akin to say superstitious behaviour in animals, i.e. expressing elaborate sequences of behaviours in relation to rewards which they ‘think’ to be contingent on the outcome, while these behaviours have no bearing to the occurrence of reward delivery.

 

The Iowa Gambling Task (IGT) in humans is a task, which deals with opposing tendencies: differences in short-term rewards between options and differences in long-term benefits of options. The short-term high rewarding options turn out to be long-term loosing options as losses in these options are quite high, while short-term low rewarding options turn out to be long-term winning as losses in these options are rather low. Subjects learn to discriminate these options through exploration and subsequent (implicit) reinforcement learning. Structures involved in this task are the same prefrontal-striatal structures that we see go awry in disordered gambling. Hence, it comes to no surprise that disordered gamblers perform poorly in this task, i.e. choose long-term poor options. It is also no wonder that some gambling-like rat models are based on the IGT.

 

Several rodent models of the IGT exist (r-IGT), based on the same conflict between short-term and long-term benefits, with food pellets as reward and either quinine pellets or delays as punishments. A critical difference however is, that differences in long-term benefit are not related to pure long-term win or loss, but rather to more or less long-term wins of pellets. In addition, it should be noted that models based on food, and hence needing deprivation in animals to motivate the animals to perform in a task, may have a different dynamics than models based on money in humans, for instance, of which they are not a priori deprived in the human IGT. However, despite the different dimensions of rewards, underlying decision-making in these different domains similar structures may operate,. While r-IGT models differ in several aspects of rewards and punishments and their delivery, they tend to converge to the same set of structures in the brain as observed to those in humans, i.e. the rodent homologues of the human prefrontal-striatal areas. As said, in itself this may not come to a surprise as they are based on the same underlying task-related problem. Yet, it should be noted that a critical difference is that humans may start to reflect on their choices or search for new information, leading to a second or additional round of processing, while in all likelihood this won’t happen in rats.

 

Interestingly both in humans and rats individual differences exist in task-performance. In rats, two different and opposing types may be found. First, rats that learn the task slowly versus rats that learn the task quickly. These differences are related to differences in the activity of structures involved in the transition from exploratory activity to directing behaviour to the best long-term options: quick in fast learners, slow in slow learners. Second, rats that consistently and increasingly choose poor options. These are not observed in all tasks, suggesting either differences in task-specific features, or differences in rat populations being tested. Whatever underlies these differences, these rats are characterized by low levels of prefrontal cortex control, especially the orbitofrontal cortex, as well as high tendency of activity in striatal areas, which are sensitive to probabilities and expectations of rewards. In other probability-related tasks such gambling-prone rats are also observed. Yet, these rats are not completely insensitive to punishments as in a task where delays are inserted before rewards, they will adjust their behaviour. This may suggest a ‘fatal’ interaction between traits and test design.

 

Concluding, rodent IGT models have translational value for the human IGT, i.e. in terms of choice behaviour and underlying brain circuits. Rodent IGT models have face-validity for gambling behaviour and may unravel processes underlying gambling behaviour. A ‘gambling-like’ phenotype has been observed whose performance may be task-dependent. But what is currently clearly missing, is the transition to, and expression of, disordered gambling, i.e. costs entailing to performance as rats do not really suffer losses as they always earn food pellets and their body weight is maintained within reasonable limits. Future studies should be directed at exploring tasks that entail such costs.

Share 0
Like 0