1. IntroductionRecently, a report showed that global e-commerce sales once again hit a record high [1]. With the further development of e-commerce, online shopping has proliferated and become an essential part of people’s daily lives. However, limited by time and space, consumers cannot directly observe products when shopping online, which inevitably brings about information asymmetry [2] and makes product quality more uncertain [3]. In such cases, consumers are more likely to use cues provided by online stores to reduce uncertainty and assess product quality [4].According to the cue utilisation theory [5], products deliver a series of cues, which can be divided into extrinsic and intrinsic cues and are commonly used by consumers to evaluate product quality [6]. Intrinsic cues regard the direct attributes of the product (e.g., product specifications), and extrinsic cues are related to indirect attributes (e.g., sales volumes). Previous studies suggest that consumers are more likely to use extrinsic cues online to evaluate product quality when they cannot physically observe the intrinsic cues of products [4]. Thus, this study employs two vital and unique extrinsic cues in online shopping: monthly sales and product ratings (for short, sales and ratings are used below). Online stores frequently use sales, which have shown to play a vital role in guiding consumers to better place an order [7]. Moreover, sales have long been reported to influence purchase intentions through the herding effect [8]. Ratings, which are the simple version of consumer reviews [9], have shown to have a significant influence on consumers’ perception of product quality [10] and ultimately impact purchase intentions to a great extent [11,12,13]. A previous study researched the impact of two online cues, sales volumes and product ratings, on consumer purchase decisions, but only headsets were used as the experimental material [9]. Commonly, products can be classified as search or experience [14,15]. Search products represent products whose attributes can be assessed objectively even before purchase using second-hand data. In contrast, experience products represent products whose attributes can only be assessed after purchase and in person [14,15,16]. The headset, which was used by Wang, Meng, Liu, Wang and Ma [9], is commonly distinguished as a search product. Previous studies highlight that consumers have different information processing procedures when purchasing search and experience products [17,18]. In other words, due to consumers’ lack of ability to assess experience product quality and reduce the risk of information asymmetry when shopping online, consumers may rely significantly more on rating information to make purchase decisions (compared with search products). Therefore, this study aimed to address the following questions: (1) do experience products influence consumer online purchase intention based on ratings and sales cues? (2) If so, what specific influence occurs? (3) What are the neural and psychological processes of experience product purchase? Building on the basis of the cue utilisation theory, the cue diagnosticity framework (shortly called framework below) proposes that when consumers are faced with multiple extrinsic cues simultaneously, they intuitively classify cues according to the perceived reliability (diagnosticity) of cues [19]. The results of the classification are noted as high-scope cues (perceived to be more reliable) and low-scope cues (perceived to be less reliable) [20]. The framework suggests that the negative (positive) inferences on perceived product quality elicited by high-scope cues pass to low-scope cues, making low-scope cues less (more) diagnostic (original mechanism) [19]. Meanwhile, some studies report that high-scope cues (no matter whether positive or negative) may continuously restrain the influence of low-scope cues on purchase decision making (alternative mechanism) [21,22]. While Wang and her colleagues’ study, using a search product, provides support for the original mechanism of the framework [9], the current study, employing experience products as the key moderator, presents an interaction between sales and ratings that supports the alternative mechanism. To address the difficulty in clarifying a vital but nuanced mechanism difference in the framework and simultaneously investigate the neural and psychological processes of experience product purchase, the current study employed event-related potentials (ERPs). Accordingly, the ERP method has the following advantages: First, people lack the ability to either predict their future behaviours or identify their internal mental states through self-reporting accurately [23]. Evidence indicates that traditional self-reporting methods might provide inaccurate data [24]. Pioneering studies also report that explicitly asking participants to self-report such internal mental states and processes leading to choices is considered to alter the outcome of their judgments [23,25]. Secondly, the ERP approach has been proven to have a high temporal accuracy in explaining the neural bases of online shopping decision-making processes [26,27]. Thirdly, as suggested by Ariely and Berns [28], and Boksem and Smidts [29], brain data are less noisy than data obtained with conventional marketing methods, and neuroscience methods can provide marketing researchers with information that cannot be obtained through traditional methods. From this perspective, the ERP method can finely unveil the psychological processes of experience product purchase at the brain level and provide precious evidence for our behavioural argument. The current study makes at least the following contributions: First, it contributes to the “online purchase” literature and cue diagnosticity framework by providing additional support to the alternative mechanism of the framework using experience products as experimental materials. Second, it provides additional verification of the ERP results and explanations of the valuable previous study [9], suggests a supplementary explanation of the LPP component, and further infers the existence of a dual process. Third, it proposes clear guiding significance for online shopping practice, especially for online retailers.

The following paper is organised as follows: First, we review the relevant literature and subsequently propose hypotheses. Secondly, we describe our ERP experimental materials and method. Then, behavioural and ERP results are presented. Finally, we discuss the key findings, theoretical contributions, managerial implications, and research limitations.

5. Discussion

In this study, we investigated consumer behaviours and the underlying brain activities correlated with the influence of sales and ratings on consumer decision making when purchasing experience products online using ERPs.

5.1. Discussion on Behavioural Hypotheses

The behavioural results indicated the significant main effects of ratings and sales, and their significant interactive effect. H1a was directly supported. The sample effect test further demonstrated that ratings significantly influenced purchase intention, whether the sales were high or low. Furthermore, from the perspective of the absolute value of purchase intention, although sales significantly affected the subjects’ purchase intention under low-rating conditions, the final average purchase intention score did not exceed 3 (on a 7-point Likert scale). The same effect also appeared under the conditions of high ratings; no matter how high or low the sales were, the average purchase intention score was above 4. This showed that the influence of sales on purchase intention was limited, and it could not reverse purchase decisions towards a particular product. In other words, rating information first determined the tendency of purchase intention (such as “>4” or “<4”). Then, the sales volume affected purchase intention to a relatively small extent in this overall range. Thus, H1b was supported. Based on these results, we argue that consumers are likely to rely more on ratings (high-scope cues) and that the effect of sales (low-scope cues) is continuously inhibited. We also obtained evidence from the ERP results that further supports our argument.

5.2. Discussion on ERP Hypotheses

At the brain level, remarkable P2, N400, and LPP amplitudes were evoked during the purchase decision process.

5.2.1. P2 HypothesisP2 is an early positive component that peaks around 200 ms after stimulus onset and has been suggested to reflect the quick and intuitive processing of stimuli [48]. As reviewed above, P2 is considered to be highly sensitive to negative stimuli [49]. We discovered the main effects of ratings and sales, which provided direct support for H2. The results showed that both low ratings and low sales could elicit greater P2 amplitudes, which suggested that both of these two factors were recognised as negative stimuli by the subjects’ intuitive processing in the experiment. These results indicated that when consumers faced the two online cues, there was an intuitive detection of negative stimuli (negative information). 5.2.2. N400 HypothesisFollowing the P2 component, N400 is considered as the reflection of emotional conflict [55,56]. The stronger the emotional conflict, the larger the amplitude of N400. In the analysis of N400, we found the main effect of ratings. However, neither the main effect of sales nor the interaction between these two factors was revealed (H3 was supported). The existing N400 results suggested that ratings were recognised as high-scope cues and significantly affected consumer decision making by continuously attenuating the effect of sales. As mentioned earlier, in the cue diagnosticity framework, a high-scope cue can either “enable” or “disable” a low-scope cue by altering its diagnosticity. Thus, the emotional conflict caused by violating the guidance of ratings was more intense, and only the main effect of ratings was significant. More specifically, low-rating conditions were considered unacceptable and stimulated strong emotional conflict. It is worth mentioning that, in real online shopping scenarios, between HS&LR and LS&HR conditions, the former is rare, but the latter is sensible, especially when a new product has been recently launched. We speculated that it was also one of the reasons why LS&HR represented a more acceptable condition for the subjects in the experiment. 5.2.3. LPP HypothesisThe LPP component is recognised as an indication of the distribution of attentional resources [68]. Previous research has further inferred its implications and has suggested its correlation with emotional arousal [46,59,60] and cognitive evaluative categorisation [26,56,65,66]. Consistently with H3, only the ratings’ main effect was found in the LPP amplitudes. However, just as the deduction in the LPP Hypothesis section, the ERPs’ weakness in clarifying “one-to-one” causal relationships is long-lasting [76]. Thus, establishing whether one of these two explanations or their combination truly represents consumer psychological processes of experience product purchase is beyond the capacity of the current research. Indeed, clarifying these two alternative explanations is not the primary goal of the current research and does not influence our most significant findings (showing support for the alternative mechanism of the cue diagnosticity framework in experience product purchase). However, we still try to draw some inferences from the results. We propose that a dual process, including an emotional process represented by arousal and a cognitive process represented by evaluative categorisation, exists in the LPP stage. From one perspective, previous studies have described the LPP as a sustained P-300-like component, which may involve a neural process similar to that of P300 [77,78]. Since P300 has been found to be sensitive to both the similarity between expected and displayed information (e.g., [67]) and emotionally significant information (e.g., self-affirming [79] or social rewards [80]), the LPP may also bear these features. This argument, combined with the fact that these two explanations share the same physiological base (attention distribution) [68], further supports that these two explanations are complementary rather than mutually exclusive. From another perspective, previous marketing studies have provided evidence of the co-existence of cognitive and emotional factors in online purchase behaviours (e.g., [81,82,83,84]). Specifically, in Verhagen and Bloemers [83] study, the so-called “think-feel-do hierarchy” of purchase intention was empirically tested in the experience product context. Thus, when in a real purchase scenario, emotionally, completing a satisfying purchase after noticing high-rating information excites personal goal achievement emotions in consumers [63], which elicits larger LPP amplitudes. Moreover, cognitively, consumers intuitively categorise products with different ratings into different groups (i.e., “acceptable group with high rating products” and “unacceptable group with low rating products”) based on similarity. A high similarity between high ratings and acceptable cues elicits larger LPP amplitudes. The LPP results were in line with those of N400, which also reflected the special attention that consumers paid to ratings. The effect of high-scope cues (ratings) significantly affected the consumers’ perception of satisfactory consumption (i.e., the generation of personal goal achievement emotions and the perception of high similarity). It also continuously inhibited (or “disabled”) the effect of low-scope cues (sales). Ultimately, only ratings strongly influenced the LPP amplitude. Furthermore, the results of both N400 and LPP additionally supported and explained why the behavioural data showed that ratings had a stronger impact on purchase decisions than sales (H1b). 5.3. General Discussion

We summarise our findings and inferences briefly as follows: In a typical experience product purchase scenario, when consumers are faced with sales and rating cues at the same time, first, they quickly and intuitively recognise low ratings and low sales as negative stimuli. Later, affected by the characteristics of experience products, the exposed cues and acceptable cues in the optimal purchase scenario preset by consumers are inconsistent, which causes strong emotional conflicts. In this stage, we speculate that consumers process ratings and sales as high-scope cues and low-scope cues, respectively. Meanwhile, due to the effect of the cue diagnosticity framework, ratings (high-scope cues) suppress the impact of sales (low-scope cues) on the decision-making process. Thus, the N400 component only shows the main effect of ratings. Finally, the effect of high-scope cues continuously exists, and different conditions of ratings (high or low) trigger different extents of personal goal achievement emotions and indicate different categorisations of products based on similarity, elicit different levels of LPP amplitudes, and guide consumers to make final purchase decisions.

5.4. Theoretical Contributions and Managerial ImplicationsThis research was carried out under the inspiration of Wang, Meng, Liu, Wang and Ma [9] study and aimed to further explore its topic. Firstly, the current research contributes to the “online purchase” literature and the cue diagnosticity framework by providing additional support to the alternative mechanism of the framework using experience products as experimental materials. Wang and her colleagues’ study [9], using only headphones (a representative search product) as the experimental material, found the interaction between ratings and sales in an additive manner, which fully supported the original mechanism of the cue diagnosticity framework. The current study, using experience products as the moderator (we did not directly test the moderator effect by comparing search to experience products in a single experiment), found a different mechanism whereby high-scope cues (regardless of their positive or negative nature) continuously inhibited the expression of the diagnosticity of low-scope cues, which gave support to the alternative mechanism of the framework. Additionally, past research has found other contexts in which the utilisation of cues follows the alternative explanation of the cue diagnosticity framework, for example, consumer reviews vs. store reputation or assurance seals in an online shopping context [21] and Web privacy assurance function vs. transaction-integrity assurance functions in an online vendor context [22]. We contributed to their marvellous findings by adding a new context that supports the alternative mechanism of the framework (i.e., ratings vs. sales in online experience product purchase).

This theoretical contribution has clear guiding significance for online shopping practice. Specifically, it reminds online retailers that they cannot narrowly focus on managing sales and rating information without considering contextual factors (e.g., the type of products sold in the store). Our study shows that ratings have a greater influence on the purchase intention of an experience product and that the effect of sales is attenuated. If ratings perform worse, sales are unlikely to play a role. From this perspective, online retailers, especially those selling experience products, should build as well as maintain a good-quality signboard in the first place and provide consumers with high-quality products and services. Otherwise, blindly emphasising sales or controlling sales regardless of ratings is meaningless in enhancing consumer purchase intention.

Secondly, unveiling the brain activities related to online experience product purchase was another important objective in the current study. Considering the small sample size (only 19 subjects were recruited) in Wang’s study [9], the present paper argues that it is necessary to re-examine the neural and psychological mechanisms found in their research study. Therefore, with a comparatively larger sample size and by keeping the experimental settings highly similar, this paper conducted research and discovered the underlying psychological mechanisms of attention distribution (P2), emotional conflicts (N400), and emotional arousal and evaluation categorisation (LPP). These findings were highly similar to those of Wang’s study [9], which further verified our results’ robustness. In addition, we also provide a supplementary explanation of the LPP component and infer the existence of a dual process with the aim to help managers and scholars to appreciate this grey area and better understand the black box of online product purchase. 5.5. Limitations and Future ResearchThe present study has several limitations, which may bring guidance for future research. First, regarding “Other Moderators”, the impact of sales and ratings on product purchase intention in scenarios such as high-value products (often hedonic products) is still unclear and remains to be further discovered. Second, regarding “Results Generalization”, because the ERP experiment has strict requirements for the experimental environment, subjects, and experimental equipment, the online purchase scenario simulated in the experiment is a reduced version of reality and is somewhat different from the real online shopping scenario. Thus, the generalization of the research findings to practice should be made with great care and scrutiny. Third, regarding “Subjects”, although our study recruited a comparatively larger sample than Wang’s study, the sample size was still small. So, we call for re-examination with a larger sample size. In addition, all the participants in the experiment were Chinese students. So, there remain open questions regarding whether the conclusions suit other cultural backgrounds, since people from different countries and nationalities may act divergently when shopping online. Thus, it could be significant for future research to experiment with these online cues from a multicultural perspective. Fourth, despite the increasing popularity of neuroscientific tools in marketing research, scholars should always be aware of their limitations. As shown in our study, one major drawback of ERPs is the interpretation of “one-to-one” causal relationships. Although the current techniques cannot directly clarify the two alternative explanations of LPP components, we hope that the current research successfully shows consumer neuroscience as a complementary link in theoretical systems assessing consumer behaviour from a holistic perspective, including behavioural, physiological, and neural aspects [76].