Introduction

In the last two decades, compulsive buying behaviour has received increasing attention in the fields of consumer research (d’Astous 1990; Dittmar 2005a; Faber and O’Guinn 1988; Hirschman 1992; Manolis and Roberts 2008; Neuner et al. 2005a; Ridgway et al. 2008; Scherhorn et al. 1990) and clinical research (Black 1996; Koran et al. 2006; Krueger 1988; Lawrence 1990; McElroy et al. 1994; Mueller et al. 2008, 2009).

O’Guinn and Faber (1989, p. 155) define compulsive buying as chronic, repetitive purchasing behaviour that becomes a primary response to negative events or feelings, is difficult to stop, and results in harmful consequences. These negative consequences are not only economic in character (debt and financial problems) but also psychological and societal (O’Guinn and Faber 1989).

Researchers generally agree that compulsive buying can take on a pathological character, such as excessive gambling, and therefore requires therapeutic treatment. This is not reflected, however, in professional protocols. To date, there is no category for compulsive buying behaviour in either the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) of the American Psychiatric Association (APA 2000), or the International Statistical Classification of Diseases and Related Health Problems (ICD-10) of the World Health Organization (WHO 2006). It has, thus, far been designated as “nothing more specific than a disruption of impulse control,” under the heading 312.30 of DSM-IV-TR and under the heading F63.8 or F63.9 of ICD-10 (Lejoyeux et al. 2000; McElroy et al. 1992; Mueller et al. 2009).

Apart from the issue of classification in the DSM and ICD, it is particularly important for consumer policy to understand the extent of this behaviour. Recent representative studies indicate prevalence rates of 8% in Austria (Kollmann and Unger 2010), 7% in Germany (Mueller et al. 2009) and 6% in the USA (Koran et al. 2006). These numbers lead to the question of possible causes.

Numerous studies have investigated the causes of compulsive buying behaviour and its connection to sociodemographic and psychological factors. Affected persons exhibit increased impulsiveness, deficits in impulse control (self-control), low self-esteem, depression, social anxiety, money management difficulties, disruption of autonomy orientation, and a greater materialistic orientation (Black et al. 1998; Dittmar 2005a, 2005b; Faber 1992, 2004; Faber and Vohs 2004; Mueller et al. 2009; Rose 2007; Scherhorn 1990; Scherhorn et al. 1990; Spinella et al. 2007). The influence of advertising, the significance of consumption in society, and the use of debit and credit cards are documented in various studies (d’Astous and Bellemare 1989; Faber and O’Guinn 1989; Faber et al. 1987; Neuner et al. 2005a, 2005b; Scherhorn et al. 1990). Regarding gender, recent studies reveal that both women and men are affected by compulsive buying behaviour. While some studies indicate that women and men are affected to the same extent (Koran et al. 2006; Mueller et al. 2009), the majority of studies concludes that women are more often and more heavily affected than men (Black 2007; Dittmar 2005a; Kollmann and Unger 2010; Manolis and Roberts 2008; Raab et al. 2005; Ridgway et al. 2008).

With these investigations as a background, this study offers an initial contribution to the understanding of the neurological correlates of compulsive buying behaviour.

The demonstration of neurological differences between non-compulsive (“normal”) and compulsive buyers in terms of the neuronal control of buying behaviour is of relevance for consumers and consumer policy in the following ways:

  1. 1.

    Public awareness and understanding of the problematic nature of this behaviour are likely to increase. This could lead to more willingness of public institutions to support research in this field and to provide resources necessary for consumer information.

  2. 2.

    It would support the enforcement of legal regulations that help to reduce the spread and severity of this behaviour. This could, for example, establish the official recognition of compulsive buying as a disease.

  3. 3.

    It could help to overcome the still dominating model of the homo oeconomicus in economic science and consumer policy. Instead, a more realistic concept of the human being might be created, representing a solid scientific basis for actions in the interest of consumers and for their protection as well.

Neural Foundations

A number of recent contributions from medicine, psychology, neuroscience, and neuroeconomics relate to the neural foundations of compulsive buying behaviour. Of relevance here is that, for compulsive buyers, the possession or use of products is not of decisive importance; instead, it is the process or act of buying that is integral to the addiction (d’Astous 1990, p. 16; O’Guinn and Faber 1989, p. 147). In general, the process of acquisition is experienced as satisfying. This positive factor, however, is set against the negative factor of the price that must be paid. Funds used for purchase represent a loss in that they cannot be used for other purposes, like saving or the purchase of other products (Prelec and Loewenstein 1998). Therefore, in learning theory, the price manifests as none other than a punishment for purchasing products, and the pain of paying plays an important role in consumer self-regulation (Prelec and Loewenstein 1998, p. 4).

The fact that purchasing decisions involve a trade-off between the pleasure of buying and the pain of paying is consistent with recent neuroscientific evidence. For instance, different studies have revealed that neuronal activity in the striatum (nucleus accumbens) correlates with self-reported positive arousal and product preferences and precedes purchase of products and investments (Ariely and Berns 2010; Erk et al. 2002; Knutson et al. 2007; 2011; Plassman et al. 2007; Peterson 2007).

Moreover, drug, nicotine, and gambling addicts showed higher activity in the nucleus accumbens (NAcc) merely from looking at a syringe, a cigarette or a gaming machine, respectively (Berridge 2003; Birbaumer and Schmidt 2006; Volkow et al. 2002; Reuter et al. 2005).

On the other hand, numerous studies reveal that anticipation of pain and/or punishment activates the insula, among other areas, and that insula activation correlates with self-reported negative arousal (Bechara and Damasio 2005; Buchel and Dolan 2000; Knutson et al. 2011; O’Doherty et al. 2001: Ploghaus et al. 1999; Paulus et al. 2003). Thus, activation of the insular regions has been hypothesized to play a critical role in loss prediction (Paulus and Stein 2006) and in the process of deciding to buy or not to buy a product (Preuschoff et al. 2006; Knutson et al. 2007).

This motivates the question of whether compulsive buyers show a higher activation of the striatum (NAcc) and a lower activation of the insula in a purchasing decision situation. This possible difference between non-compulsive and compulsive buyers should also result in a lower activity of the ventromedial prefrontal cortex (vmPFC). The vmPFC generally plays an important role in planning and executing actions (Bechara et al. 1994; Bechara and Damasio 2005; Clark et al. 2008; Kringelbach 2005).

People suffering from damage of the vmPFC often show poor judgement and high impulsiveness, and tend to block out negative consequences of their decisions (Bechara et al. 1994, 2000; Berlin et al. 2004; Clark et al. 2008; Damasio 1994). These deficits represent the lack of a crucial mechanism required for controlled decision making.

Based on these findings we speculate that the striatum (NAcc), the insular and the prefrontal cortices (vmPFC) will show differences in activity between non-compulsive and compulsive buyers during the buying process. Based on the experimental paradigm of Knutson et al. (2007), the simulated buying process is divided into three phases: (1) the presentation of products, (2) products and prices, and (3) products and price including the purchase decision.

Specifically, we propose the following main hypotheses:

  1. (H1)

    During phase 1, compulsive buyers will reveal a significantly higher activity in the ventral striatum (NAcc) compared with non-compulsive buyers.

  2. (H2)

    During phase 2, compulsive buyers will show a significantly lower activity in the insular cortex compared with non-compulsive buyers.

  3. (H3)

    During phase 3, compulsive buyers will reveal significantly lower activity in the vmPFC compared with non-compulsive buyers.

Methods

The study is divided into four steps: (1) Selecting and recruiting subjects; (2) conducting the study using functional magnetic resonance imaging (fMRI); (3) examining the participants using a questionnaire; and (4) analysing the study data.

Subjects (Step 1)

In total, 49 women took part in the study, 23 of whom were compulsive buyers and 26 of whom were non-compulsive buyers. The limitation to female participants is based on the following two thoughts: (1) Most studies regarding compulsive buying behaviour indicate that women are more intensely and more often afflicted with compulsive buying than men, even given more recent studies and possible methodological artefacts (Black 2007; Dittmar 2005a; Kollmann and Unger 2010; Manolis and Roberts 2008; Raab et al. 2005; Ridgway et al. 2008); (2) there are differences between men and women with regard to products they prefer to buy (Black 2001; Christenson et al. 1994; Mueller et al. 2008; Scherhorn et al. 1990). Taking the influence of the gender factor into consideration would have led to a corresponding increase in test participants and additional provision of stimulus material for the fMRI study.

Compulsive Buyers (Experimental Group)

Test participants for the experimental group (compulsive buyers) were found via announcement of the planned study in various internet forums, by initiating contact with psychotherapists and self-help groups related to compulsive buying, and in the context of interviews dealing with the topic of compulsive buying in various types of media (the press, radio, or television). A crucial requirement for the selection of the actual participants from all the interested individuals was that they had to be undergoing psychotherapeutic treatment with a psychiatrist or psychologist due to their buying behaviour. Thus, the formation of the experimental group of compulsive buyers was based on an objective external criterion. Furthermore, the participants had to fill out the German Compulsive Buying Scale and achieve a score of 45 or more (Raab et al. 2005; Scherhorn et al. 1990). This scale is an adapted and modified version of the compulsive buying scale from d’Astous (1990) and is accepted as the standard procedure for measuring compulsive buying behaviour in research and clinical practice of German-speaking countries (Germany, Switzerland, and Austria) (Glaesmer and Singer 2008; Mueller et al. 2008).

The instrument includes 16 items (Raab et al. 2005; Scherhorn et al. 1990). For each item, the extent of agreement or disagreement is expressed on a four-point Likert scale, ranging from “I don’t agree” (1) to “I totally agree” (4). The instrument had previously been tested successfully in several surveys in terms of its validity and reliability (Raab et al. 2005; Scherhorn et al. 1992). Following Faber and O’Guinn (1989, 1992), respondents were classified as “compulsive buyers” if they were at least two standard deviations above the mean value on the German Compulsive Buying Scale (M = 26.46; SD = 9.06). This means that consumers are classified as being “compulsive” in terms of the German Compulsive Buying Scale when they reach a score of 45 or more. This score accords very closely with the score of 26 self-selected West German compulsive buyers (M = 45.23; SD = 10.10) who were screened in 1989 using in-depth interviews and psychological instruments such as self-esteem, psychasthenia, depression, negative feelings, and locus of control (Raab et al. 2005; Scherhorn et al. 1990). All individuals who had entered psychotherapeutic treatment due to their buying behaviour achieved a rating on the scale exceeding the cut-off score of over 44 (M = 53.57; SD = 3.84). Additionally, the sociodemographic characteristics of age and income were collected.

Non-compulsive Buyers (Control Group)

After the selection of the experimental group (compulsive buyers), the control group (non-compulsive buyers) was recruited via an advertisement in a regional newspaper. Applicants were classified as non-compulsive buyers if (1) they were not in psychotherapeutic treatment on account of their buying behaviour and (2) they achieved a score of less than 36 on the German Compulsive Buying Scale. A score of less than 36 characterizes inconspicuous (non-compulsive) buying behaviour (Raab et al. 2005; Scherhorn et al. 1990). None of the participants in the control group achieved a score over 31 (M = 21.50; SD = 4.13). Additionally, the sociodemographic characteristics of age and income were recorded. To control for the influence of age and income, the control group members were matched on these variables (see Table 1).

Table 1 Characteristics of the compulsive and non-compulsive participants
Full size table

fMRI Study (Step 2)

In order to test our hypotheses, we chose fMRI instead of other techniques such as electroencephalography (EEG) or magnetoencephalography (MEG), because the fMRI has a better ability to investigate subcortical structures. The expected involvement of the striatum (nucleus accumbens) could not have been investigated sufficiently with the use of EEG or MEG (Babiloni et al. 2009).

fMRI Task

Participants were scanned while engaging in an adapted Saving Holdings or Purchase task (Knutson et al. 2007), which consisted of a series of trials, identical in temporal structure with jittered inter-trial intervals, in which participants could purchase products (see Fig. 1). Participants viewed a labelled product (4 s); viewed the product’s price (4 s); chose either to purchase the product or not by selecting yes or no, as presented randomly on the right or left side of the screen; then fixated on a cross-hair (2 s) prior to the start of the next trial.

Fig. 1
figure 1

Example picture sequence of the simulated buying process

Full size image

A total of 100 products were presented in this manner to each participant. The price of the products varied from 1 to 50 Euros. To simulate a buying situation as realistically as possible, each participant received the equivalent of 50 Euros at the outset of the fMRI study. They could opt to spend or not spend this amount on the presented products. At the end of the study procedure, the participant received one of the purchased products, selected randomly. Any unspent funds were paid out to the participant in cash.

Through the random selection of only one purchased product after the fMRI examination, the purchase decision for each product was always based on the total disposable amount of 50 Euros. Before the beginning of the fMRI study, each participant was informed of this procedure and questioned as to whether they had understood the instructions. All questions of the test participants were answered by the investigators.

The selection of the 100 products was made according to the following three criteria: (1) The products were selected from six product categories (accessories, drinks, clothing, cosmetics, jewellery, and sweets). These categories of products were chosen because they included the preferentially purchased products of female compulsive buyers according to previous studies (Black 2001, 2007; Christenson et al. 1994; García Ureta 2007; Scherhorn et al. 1990; Schlosser et al. 1994); (2) Due to the limitations of the study budget, the maximum price of a product was limited to 50 Euros. In fact, over 90% of monetary transactions of German consumers are in the price range between 1 cent and 50 Euros per payment (Deutsche Bundesbank 2009. p. 48); (3) For the study, 100 products were randomly selected from a retailer listing of 500 of the most sold products within the 50 Euro price range. Thus, product selection was based on real sales numbers and real purchase prices, thereby providing a high degree of realism to the simulation. The 100 selected products were photographed and presented in a random order within the study, as described above.

Scanning Procedure

Scanning was performed in a 1.5T Avanto Scanner (Siemens, Erlangen, Germany) using a TIM eight-channel head coil. During the experiment, ∼830 echo planar imaging (EPI) scans were acquired. Forty-eight slices covered the whole brain, including the cerebellum (slice orientation, C->T30). Scan parameters were: slice thickness, 3 mm; interslice gap, 0.66 mm; matrix size, 64 × 64; field of view, 195 × 195 mm; echo time, 50 ms; and repetition time, 2.91 s. The task was presented via video goggles (Nordic NeuroLab, Bergen, Norway) using E-prime presentation software (Psychology Software Tools; www.pstnet.com).

fMRI Data Analysis

The fMRI data analysis was performed using Statistical Parametric Mapping 5 (SPM5; www.fil.ion.ucl.ac.uk/spm/). Pre-processing included slice timing, realignment with unwarping, normalisation to an EPI template (re-sampled voxel size after normalisation, 3 × 3 × 3 mm) and smoothing with an 8-mm Gaussian kernel. Six vectors of onset times (product, price, and decision (for bought and non-bought decisions)) were specified. In order to model the BOLD time course in each voxel, these onset vectors were convolved with the SPM5 canonical hemodynamic response function and its temporal derivative. For each participant, parameter images of the contrasts of each condition were generated. These images were then subjected to a second-level random effects analysis using Full Factorial Design with between-participant factor of group (compulsive buyer, control) and within-participant factor buying decision (i.e., subsequently bought vs. subsequently not-bought) as a model for each phase (i.e., product, price, decision) separately. An inclusion threshold of p < .001 uncorrected with an extent threshold of at least ten contiguous voxels was applied. The Anatomical Automated Labelling Tool for SPM (Tzourio-Mazoyer et al. 2002) was used to label the clusters.

Post-Scan Analysis (Step 3)

After the fMRI examination, the participants were interviewed. The intention of the interview was to double-check previous findings regarding the differences between compulsive and non-compulsive buyers related to self-esteem, self-control, and depression. The survey encompassed the following scales:

Self-esteem

Self-esteem was measured in the study with a ten-item self-esteem scale (Deusinger 1986) with an internal consistency reliability of 0.92. This scale has been used in various compulsive buying studies in Germany (Scherhorn et al. 1990; Raab et al. 2005).

Self-control

Self-control was measured in the study with a German version of the 36-item Self-Control Scale from Tangney et al. (2004), with an internal consistency reliability of 0.86. The German version was developed in conformance with the present study on the basis of a pre-test.

Depression

The extent of depression was measured using the German version of the Basic Depression Scale of the Minnesota Multiphasic Personality Inventory-2 (Engel 2000). The internal consistency reliability for this 57-item scale rated at 0.80.

Results

During the product phase, we observed a significantly higher activity in the striatum (nucleus accumbens) for subsequently bought vs. not-bought products in both groups. As predicted in hypothesis 1 this effect was significantly higher in the compulsive buyers’ group than in the non-compulsive buyers’ group (see Fig. 2). Furthermore, regardless of whether a displayed product was bought or not, in all 100 of the products presented, higher activity of the nucleus accumbens could be observed in the group of compulsive buyers.

Fig. 2
figure 2

Activation during the product presentation phase. Activation for subsequently bought vs. subsequently not-bought products during the product presentation phase and differences in activity between non-compulsive (grey) and compulsive buyers (black) in the striatum (nucleus accumbens; p < .001; voxel threshold, 10). Results based on group comparison

Full size image

During the price phase, we observed a significantly stronger activation in the insula and the anterior cingulate cortex for subsequently not-bought products when compared to those subsequently bought. This was true for both groups of participants with significantly stronger insula and cingulate cortex activity in the non-compulsive buyers’ group. The significantly lower insula activity in the compulsive buyers’ group supports hypothesis 2 (see Fig. 3).

Fig. 3
figure 3

Activation during the price phase. Stronger activation during the price phase for subsequently not-bought products compared with products subsequently bought and differences in activity between non-compulsive (grey) and compulsive buyers (black) in the insula (p < .001; voxel threshold, 10). Results based on group comparison

Full size image

In the decision phase, we found stronger activity in the vmPFC and the dorsolateral prefrontal cortex for bought products vs. not-bought products. We were unable to substantiate a lower activity in the prefrontal cortex of compulsive buyers, as formulated in hypothesis 3. The only significant difference between the two groups was reflected in anterior cingulate cortex activity. We observed significantly higher activity especially for bought products in the compulsive buyers’ group (see Fig. 4). Moreover, they purchased significantly more products (M com = 34.87, M noncom = 7.58; t = 7.77, df = 29.7; p < .001) spent a higher total average amount (M com = 923.64, M noncom = 166.81; t = 7.15; df = 24.5; p < .001), and also increased their average expenditure per purchased product, which was higher than that of the non-compulsive buyers (M com = 25.67, M noncom = 16.89; t = 3.61; df = 36.1; p < .001).

Fig. 4
figure 4

Activation during the decision phase. Stronger activity for compulsive buyers than for non-compulsive buyers during the decision phase in the anterior cingulated cortex (p < .001; voxel threshold, 10). Results based on group comparison

Full size image

The post-fMRI scan analysis yielded the following results: Compulsive buyers possess a significantly lower measure of self-esteem (M com = 32.13, M noncom = 47.92; t = −5.66; df = 47; p < .001) and ability to control themselves (M com = 104.39, M noncom = 122.08; t = −3.82; df = 47; p < .001). They also exhibited significantly higher depression values than non-compulsive buyers (M com = 30.26, M noncom = 22.35; t = 4.35; df = 47; p < .001).

Conclusions and Discussion

The intention of this study was to analyse for the first time the neural correlates of compulsive buying. The results reveal significant differences between non-compulsive (“normal”) and compulsive buyers in various brain regions. The hypothesis that compulsive buyers show a higher activity in the ventral striatum (nucleus accumbens) during the presentation of purchasable products was supported. The hypothesis that compulsive buyers show a lower activation of the insula during the presentation of product and price than non-compulsive buyers was also supported. The results are consistent with other neurological studies of buying behaviour. Higher activity in the nucleus accumbens is related to positive arousal, the preference of a product, the wanting of a product and the purchasing probability (Knutson et al. 2007; Linder et al. 2010). Insular activity on the other hand was found to be associated with negative arousal and related to the purchasing probability. It increased with the price subjects had to pay for the product, indicating price sensitivity (Knutson et al. 2007). The observed differences could be an explanation for the fact that compulsive buyers lose control over their buying behaviour.

In contrast to our hypotheses, we did not find evidence for differential prefrontal activation between compulsive and non-compulsive buyers in the phase involving decision for or against purchase of presented products and their price. We observed significantly higher anterior cingulate cortex activity in compulsive buyers. According to the current state of knowledge, the anterior cingulate cortex is especially active in evaluating the degree of conflict in decisions connected with winning and losing (Bush et al. 2002; Pochon et al. 2008). It is also noteworthy that higher anterior cingulate cortex activity has been detected in depressed persons (Knutson et al. 2007; Mayberg et al. 2005). This is meaningful since various studies indicate that compulsive buyers are dealing with depression (Lejoyeux et al. 1997; 2000; Mueller et al. 2008; Scherhorn et al. 1990; Schlosser et al. 1994). The group of compulsive buyers that was examined also showed significant symptoms connected with depression.

The present results can only be one first step towards a deeper understanding of the neurological correlates of compulsive buying. This is due to the restricted number of examined subjects, the limitation of the study to female buyers and the overall design of the study. In spite of the limitations of this study, its results and further research in this field could help to strengthen the awareness for compulsive buying in society and consumer policy. This could lead to more willingness to support research in this field and to provide resources necessary for consumer information and preventive actions. Moreover, the results could also help to overcome the still dominating model of the homo oeconomicus and to create a more realistic concept of the human being in economic decision making. Taking into account all eligible criticism, neurological findings and the corresponding methods are likely to become an integrated and accepted part of consumer research and policy (Hubert 2010, p. 816). The integration of neuroscience into consumer policy is in the interest of the consumer and helps to foster consumer protection (Kenning and Linzmajer 2010; Raab et al. 2010).