Several months ago, I posted a study on videogame addiction and its relation to cognitive bias towards videogame words. Following Decker & Gay’s publication, two similar studies were published during the past summer of which they made references to Decker & Gay (2011). Not wanting to leave out those two studies, I decided to review all three in one post.
There is considerable dispute regarding the nature of excessive or problematic Internet-related behaviour and whether it constitutes a clinical addiction. Classification of excessive gaming is hindered by a lack of experimental research investigating behavioural responses from gamers and comparing these patterns to those found in established addictions. We investigated whether an attentional bias for gaming-related words existed for addicted Massively Multiplayer Online Role-Playing Gamers (MMORPGers) identified using the Addiction–Engagement Questionnaire. Forty frequent MMORPGers (15 female) and 19 non-MMORPGers (eight female) completed a computerised modified Stroop task comprised of game-related, negative and neutral word lists, Addiction–Engagement Questionnaire, Depression, Anxiety and Stress Scale 21, gaming-related variables. The results indicated that addicted MMORPGers had significantly longer reaction times to negative and MMORPG words compared to neutral words, whereas highly engaged and non-MMORPG participants showed no such bias. The presence of an attentional bias in addicted MMORPGers is comparable with research investigating this behavioural response in established addictions.
The aim of this study was to examine whether behavioral tendencies commonly related to addictive behaviors are also related to problematic computer and video game playing in adolescents. The study of attentional bias and response inhibition, characteristic for addictive disorders, is relevant to the ongoing discussion on whether problematic gaming should be classified as an addictive disorder. We tested the relation between self-reported levels of problem gaming and two behavioral domains: attentional bias and response inhibition. Ninety-two male adolescents performed two attentional bias tasks (addiction-Stroop, dot-probe) and a behavioral inhibition task (go/no-go). Self-reported problem gaming was measured by the game addiction scale, based on the Diagnostic and Statistical Manual of Mental Disorders-fourth edition criteria for pathological gambling and time spent on computer and/or video games. Male adolescents with higher levels of self-reported problem gaming displayed signs of error-related attentional bias to game cues. Higher levels of problem gaming were also related to more errors on response inhibition, but only when game cues were presented. These findings are in line with the findings of attentional bias reported in clinically recognized addictive disorders, such as substance dependence and pathological gambling, and contribute to the discussion on the proposed concept of “Addiction and Related Disorders” (which may include non–substance-related addictive behaviors) in the Diagnostic and Statistical Manual of Mental Disorders-fourth edition.
It’s Halloween already?! I just barely got warmed up to my classes and we’re now half-way through the quarter!
The three studies share a common theoretical background in that they are interested whether videogame addiction shares a psychological phenomenon previously shown in other behavioural addictions (e.g. gambling). The phenomenon in question is attentional bias, it is a process where cues related to the target addiction (i.e. words, images, sounds, smells) captures the individual’s attention that is implicit and/or automatic. This attentional bias is observed through reaction times and accuracy where addicts respond differently to addiction-related cues than controls.
In addition to attentional bias, the studies examined this phenomenon in different aspects. Metcalf & Pammer (2011) examined differences between highly engaged, but not addicted players versus addicted players as previously explored by Charlton and Danforth (2007). Van Holst et al. (2011) examined attentional bias’ relationship with diminished self-regulation, which would manifest in diminished inhibitory responses.
Decker & Gay: 42 undergraduate students, average age is 21 years, all male. They have 12 WoW players and 30 non-players.
Metcalf & Pammer: 60 undergraduate students, average age is 21 years, 23 females and 37 males. They have 40 (15 female) MMO players and 20 (8 female) non-MMO players.
Van Holst et al.: 92 male Dutch high school students, average age is 15.1 years (SD = 1.27), all male. It seems they examined a more general sample rather than a focused sample like the previous two.
Common questionnaires across the studies are demographical and gameplay time questionnaires. However, differences appeared in that van Holst et al. (2011) asked participants’ weekly time, days per week and hours per day. Metcalf and Pammer (2011) asked participants their frequency of MMO-play per week in the past month, and length of time playing. Gay & Decker (2011) asked weekly hours of playtime.
Decker & Gay (2011) adapted two addiction measures previously used for alcoholism, called the AUDIT and T-ACE questionnaires.
Metcalf & Pammer (2011) used Charlton’s (2002) Addiction-Engagment questionnaire, a 24-item 7-point agreement scale assessing for measures of engagement and addiction.
Van Holst et al. (2011) used Lemmens et al.’s (2009) Game Addiction Scale, a 21-item 5-point frequency scale assessing for seven criteria of pathological gaming: salience, tolerance, mood modification, relapse, withdrawal, conflict and problems.
Decker and Gay (2011) used the Back Depression Inventory, a 21-item commonly used scale .
Metcalf & Pammer (2011) used the Depression, Anxiety and Stress scale, a 21-item 4-point “apply to me” scale, another commonly used scale.
Attentional bias and response inhibition measures
Decker & Gay (2011) used the go/no-go task where participants are asked to respond by pressing a key to target stimuli from distracter stimuli. They used 26 WoW-related words, 11 are positively valenced (e.g. Level up) and 15 are negatively-valenced (e.g. Nerf) and 33 English words. The dependent variables are reaction time, discrimination and disinhibition.
Metcalf & Pammer (2011) used the stroop task where participants are asked to press a button corresponding to the colour of the word that was shown on the screen. They used 20 videogame-related words taken from Nick Yee’s Daedalus Lexicon and from several MMO-related websites (probably from their glossary) and 20 negatively-valenced words (e.g. angry) and 20 neutral words. The dependent variable is the error and response latency.
Van Holst et al. (2011) used two attentional bias measures. The first is the dot-probe task where participants were shown two pictures on two sides of the screen, one random side would show a screenshot from a videogame and the other side would show matched neutral animation picture. There are 50 picture pairs, the game screenshots are taken from 18 videogames that were picked by participants of another study. The dependent variables are reaction time bias and error bias.
Their second measure is the stroop task where participants are asked to press a button corresponding to the colour of the word that was shown on the screen. They used 17 game-related words and 17 non-gaming words. The dependent variables are error bias and reaction time bias.
Finally, Van Holst et al. (2011) used the go/no-go task, similar to Decker and Gay’s in that the participants were asked to respond to pictures of cars and not to respond to pictures of videogames. They used 120 pictures of cars and 40 pictures of videogames. The dependent variable is the number of responses to videogames.
Decker & Gay reported that the WoW players displayed faster overall reaction times, in particular towards positive valenced WoW words. They were also more reactive towards the WoW words, i.e. response bias towards the jargon.
Metcalf & Palmer reported no overall differences in error responses between the three groups. They also reported no overall differences in response latency between the three groups, but they reported a main effect to word types. They did found significant interaction effects between the word types and groups, specifically among the participants in the addicted group took longer to react to both negative and gaming words than neutral words. The other groups showed no differences in latency to the word types.
Metcalf & Palmer found a significant negative correlation between the addiction score and the interference scores (calculated by subtracting the mean reaction times for neutral words from the negative and gaming words). The same is somewhat true (when removing an outlier) for the Depression, Anxiety and Stress.
Van Holst et al. separated the adolescents into three groups according to the range of their game addiction scale (from 1 to 3.43). They separated them to the percentiles, 33th and 67th percentile. So, the low group have average scores of 1.00 – 1.86 (N=31), the medium problem have average scores of 1.86 – 2.33 (N=31) and the high problem group have average scores of 2.33 – 3.43 (N=30). So, bear in mind about this statistical categorization of adolescents, but they have found a positive correlation between the addiction score and the number of hours of videogame play.
Dot-probe task results indicated there were no differences in reaction time bias, but there was a significant positive correlation between addiction score and error bias. Those with higher addiction scores made more errors to gaming pictures, thus indicating an attentional bias to gaming pictures.
Stroop task error-bias scores towards game-related words were positively correlated with addiction scores. However, there were no correlations for reaction time bias scores.
In the go/no-go task, addiction scores were positively correlated to the number of failed no-go trials.
Decker & Gay studied only male WoW player participants, the order of words presented during the experimental task might be an issue, but I am not following it quite right.
Metcalf & Palmer studied participants who played MMOs, so this is a generalizability issue, but also a case of co-morbidity (poorly borrowing a term from clinical psychology) in that how can you find MMO player who don’t play single-player or other types of videogames? It is possible you can find some participants who never played MMOs, but this is a sampling issue here. The authors argued that the salience of the words used in the study may not affect other types of gamers (i.e. those I mentioned earlier).
Van Holst et al. cited using a convenience sampling technique with their adolescent population, so another issue of generalizability and representativeness. They reported that a floor effect might explain the non-significant results for their reaction time variables in that the participants were very fast.
The take home message from those three studies have found some validating evidence that videogame addiction, actually online videogame addiction just to be specific, have cognitive effects similar to those found in other behavioural addiction, like gambling. Nevertheless, all study authors argued that much work is to be done to construct a theoretical model to explain videogame addiction and attentional bias is, but one identified conceptual block.
Decker, S. A., & Gay, J. N. (2011). Cognitive-bias toward gaming-related words and disinhibition in world of warcraft gamers. Computers in Human Behavior , 27 (2), 798-810. DOI: 10.1016/j.chb.2010.11.005
Metcalf, O., & Pammer, K. (2011). Attentional bias in excessive massively multiplayer online role-playing gamers using a modified stroop task. Computers in Human Behavior , 27 (5), 1942-1947. DOI:10.1016/j.chb.2011.05.001
van Holst, R. J., Lemmens, J. S., Valkenburg, P. M., Peter, J., Veltman, D. J., & Goudriaan, A. E. (2011). Attentional bias and disinhibition toward gaming cues are related to problem gaming in male adolescents. Journal of Adolescent Health . DOI: 10.1016/j.jadohealth.2011.07.006