1 Introduction

Online transactions have become easier for consumers as more applications are available in the marketplace. According to the statistics reported by the comScore MMX Multi-Platform [7], there have been more mobile device users than desktop computer users in Taiwan since May of 2016 in the age group of 25 to 54 years. Using a mobile device to do an online transaction is similar to doing a traditional one on e-commerce websites where consumers still need to go through a series of complex purchasing decision-making processes [5, 10, 24]. From the perspective of the S-O-R (stimulation-organism-response) psychological model, consumers’ purchasing decision processes can be systematically presented [16, 31]. Even though consumers save time on completing a transaction by a few clicks via their mobile devices, due to the significant effects of information asymmetry on online transactions, consumers may perceive high risks of financial loss and encounter fraudulent methods, such as low product quality, transaction interception, product price and function misrepresentation, seller evaluation inflation and shill, and delivery failure [8, 12, 21].

Online consumer research on trust behavior has pointed out that online sellers not only should be established as a secure and privacy-oriented website, but also should consider the establishment of credibility, so that consumers feel that the site is trustworthy [6]. However, online fraud has recently become prominent in online transactions, coupled with higher risks involving online auction platforms [12]. On the platforms, continuous or simultaneous types of online auctions are both available to consumers [17]. Sellers can display more than one product item as a combo in one auction, and buyers are able to maintain a degree of anonymity while purchasing products on online auction platforms [19]. Therefore, both the consumers and the sellers often lack information about the counterparty [13]. This anonymous mechanism makes people more opportunistic and increases online market opportunistic behavior [15].

The brain can be divided into different regions, and each region controls behaviors in different ways, including cognition, emotion, decision making, calculation, and memory [23]. Online consumers’ purchasing intention and its influential factors, such as purchasing habits, can also be forecasted by analyzing time series brainwave data [22, 25]. When an α wave ranging from 8 to 13 Hz is detected in a person’s left frontal lobe, it means that he/she is having a positive emotion and is able to actively take action. Contrarily, a negative emotion in a passive reaction is correlated with brain neural activation in the right frontal lobe [26]. An increasing number of neuromarketing studies has provided empirical evidence that a non-intrusive approach to retrieving consumers’ brainwave data is the most objective way to understand consumers’ purchasing decision-making processes [33]. In the e-commerce context, the most direct way to best satisfy consumer needs is to detect consumers’ active brain neurons, which represent the consumers’ thoughts regarding purchasing preferences and product price [2]. For example, electroencephalography (EEG) is one of the practical ways to retrieve consumers’ brainwave data.

The present study was framed by the Engel-Kollat-Blackwell (EKB) model and utility theory and adopts the Stimulus-Organism-Response (S-O-R) psychology model, to investigate the effect of different feedback mechanisms on consumer final purchasing decision-making in an online auction purchase process. The purpose of this study was to explore how different transaction risk scenarios on auction platforms can significantly and differently influence online consumers’ brainwaves.

2 Literature Review

Consumer purchasing decision-making process refers to all the activities that consumers engage in to meet their needs or wants. Such activities include search, choice, purchase, use, and product evaluation. The process consists of two aspects: the subjective psychological events and objective material-based events. The behaviors displayed during the process are called purchase behaviors [21]. Consumer purchasing decision encompasses the process in which consumers carefully evaluate a product itself as well as its brand value and characteristics, choose, and purchase a product in order to meet certain needs [31]. In other words, to meet their needs, consumers have a purchase motivation, then they analyze and evaluate purchase options, choose one of the many options, and finally decide to execute the optimal purchase option, all the while conducting a systematic decision process from start to finish. Consumer purchase behavior consist of the following steps [9]: problem recognition, search for information, alternative evaluation, choice, and outcome evaluation. The most important of these steps is the choice being made because choices are the end products of complex cognitive activities of human brains, and the accuracy of a choice directly determines whether and how a purchase behavior will occur [31].

2.1 Theory of Consumer Behavior

Decisions made during consumer purchasing process include psychological and cognitive events, as well as the actual final purchase behaviors which are affected by endogenous and exogenous factors [4, 27]. Therefore, the process of purchase behavior – meeting one’s needs through evaluating and choosing products – is a process of evaluation and decision [9]. The EKB model assumes that after obtaining sufficient information, consumers may generate several options, evaluate those options, and make a choice. The standards for evaluating option are formed by the consumers’ purchase perspectives and outcome expectations, and these standards present themselves as product characteristics preferences. Based on the EKB theoretical model, consumers of digital transactions have a vastly different degree of autonomy and control in their purchase process compared to the purchase behaviors found in traditional stores. Online purchase provides more options in less time, increases purchase speed, and offers innovative post-purchase services [18].

In addition, analyzing consumer behaviors quantitatively enables one to use needs theories to investigate the relationship between desires and utility [34]. Desire is the feeling of not being able to have something, and it has no limits. When a desire is satisfied, another rises. Utility is the sense of satisfaction gained when purchasing a product; it is a subjective feeling that shifts constantly due to time, location, and human factors. According to consumer behavior theories and needs theories, when price is fixed, consumer surplus increases, which determines the price consumers are willing to pay. In other words, as product quantity increases, marginal utility decreases, which lowers the price consumers are willing to pay. Similarly, the larger the utility, the higher the price that consumers are willing to pay. From a psychology perspective, one can apply the S-O-R model to examine consumer purchase behavior. S-O-R stands for stimulation-organism-response. The theory states that consumer purchase behaviors start with motivations formed by the synergy of psychological factors and external stimuli, then proceed to product purchase decisions, actual act of purchase, and finally, post-purchase evaluation [16, 31].

2.2 Fraudulent Online Transactions

Online shopping touts efficient and comfortable shopping experiences, breaking the tradition of physically going out to shop. However, scammers use a variety of methods to conduct illegal transactions online. Especially, when consumers use private communication software to shop, because the platforms do not maintain transaction records, the Q&A tools in such platforms do not count as evidence for completed transactions, making the platform owners unable to assist with handling complaints when conflicts arise or when consumers are scammed as result of pursuing alleged perks [28]. Generally, concerns regarding online transactions are rooted in: (a) the lack of face-to-face interactions between sellers and buyers causes their inability to confirm each other’s identities; (b) the steps for completing an online purchase are complex, making it possible for payment information to be leaked and increasing the odds of it being stolen; and (c) online purchases require submitting personal information and credit card numbers at the risk of these being stolen. Kalakota and Whiston [14] pointed out that online transactions have three primary security requirements: privacy, confidentiality, and integrity. Thus, encryption, digital signature, public key infrastructure, key logger, etc., are effective tools to increase online transaction security.

2.3 Online Consumer’s Brain Neural Circuits

The human brain has several regions that are associated with decision-making and cognition, including emotions and socialization. These regions’ nerves are activated when humans engage in online auctions or e-commerce. For example, the prefrontal cortex and the anterior cingulate gyrus are related to decision-making. If there is uncertainty during a decision-making process, the orbital frontal cortex and parietal cortex become activated. Additionally, the human emotions of happiness and pleasure are managed by three brain regions: nucleus accumbens, anterior cingulate cortex, and putamen. The unpleasant emotions are also managed by three regions: superior temporal gyrus, amygdaloid, and insula cortex. Morin [26] found that detecting the α brainwave (8–13 Hz) produced by the left frontal lobe indicates positive emotions. By inference, such brain activity is a good predictor of being motivated to take action [20]. The right frontal lobe is generally associated with negative emotions, which often makes a person withdraw from taking action. We now briefly describe the electroencephalography (EEG), an instrument for detecting brainwaves, before proceeding to discuss neuralmarketing research findings that involved using this instrument.

2.4 The Application of EEG in Neuromarketing Research

EEG is a non-invasive technology that measures brain activity. It can detect tiny brainwaves under the scalp and does not harm the human body. It converts electrical signal into numbers and sends the data through an amplifier, then displays the time series of the voltage values. It can show the changes that a brain goes through in different conditions or states, so it is a valuable tool for providing data for analysis and research [33]. The greatest advantage of EEG is its time sensitivity (recording up to 2,000 data points per second), which enables it to capture the fast-changing brain activities and processes such as awareness, recognition, and emotion exercises, better than other brain imaging tools like MRI or PET scan [29].

EEG’s application in the field of neuromarketing allows economists to examine the brain’s processes that drive consumer decisions. In neuromarketing research, often the research participants are invited to observe especially designed experimental materials, and researchers obtain the participants’ brainwave’s time series graphs in order to understand what factors in the purchase process have impact on consumer’s purchase intention [22]. Furthermore, common purchase habits or patterns can be identified, so that key elements can be modified to more closely conform to consumer behaviors [25]. One example is the text configuration on merchandizes and webpages. Because the human brain prefers images over texts, consumers may adjust the proportion and placement of images and texts to improve the brain’s ability to process information, to increase the purchase rate of online shoppers and e-commerce users, and to improve consumers’ preference of and attention to the merchandizes’ characteristics and similar products [26]. Some research studies use mobile lab instruments. They ask participants to explore real or virtual stores and measure the individuals’ mental statuses as they purchase products or services, and thus obtain consumers’ purchase habits and decision-making process in real-world settings [1]. Neuromarketing can also be applied when investigating consumer preference of brands and advertisement [22]. Price is a major factor in influencing purchase decisions; it can even be used as a tactic to attract buyers. Neuromarketing research can be used to discover areas in the brain that help consumers rationalize their preferences and to induce the most profitable response when stimulated by the highest price [2].

3 Methodologies

The research framework is sketched in the figure below (Fig. 1). The stimulation (S) in the study consisted of the two online transaction feedback mechanisms for displaying product information: one or two products to purchase and with-without textual transaction labels. The organism (O) had two dimensions: consumer background (in this case, college major as either engineering or liberal arts) and consumer purchase needs (one item or a two-item combo purchase). The response (R) was the consumer purchase decision, both in terms of the efficiency in making the decision and in terms of the accuracy of that decision. The entire research process utilized EEG to generate consumer brainwave graphs, and the efficiency and accuracy of purchase decisions were recorded simultaneously.

Fig. 1.
figure 1

Research framework.

Full size image

Based on prior research, five brain regions were the focus of the study: Fz (frontal lobe, focusing on the orbital frontal lobe cortex), FCz (between frontal lobe and parietal lobe, focusing on the anterior cingular gyrus cortex), Pz (occipital lobe cortex), Cz (parietal lobe cortex), and CPz (between occipital lobe and parietal lobe) [3, 29, 30, 35]. Of the five regions, the frontal lobe (Fz and FCz) is located near the forehead and is related to higher cognition (working memory, thinking, judgment, planning, creating), language, personality, and exercise. The parietal lobe (Cz and CPz) is located at the back of the head. It acts as the processing center for sensory information and is responsible for high-level perceptions such as visual-spatial ability, attention, touch, pressure, and pain. The occipital lobe (Pz) is located at the center of the back of the brain and is the simplest region, primarily responsible for vision (shape, size, color) and responds to changes in number size. When purchasing product quantities larger than available items, this area is expected to be stimulated and will show distinctive amplitudes in the waves. In this study, participants were required to observe the images in the experiment that show the number of items available and the number of items desired. When the number to be purchased was greater than the number of items available, the participants must make a purchase decision in a timely manner. That is, the participants’ Fz, FCz, Pz, Cz, and CPz regions would experience changes in brainwave voltage, and the EEG graphs would show distinctive amplitudes. It is worth noting that the experiment images were limited to consumer electronics, computers, communication devices, and their peripheral products for bidding. The price level matched the market value, and the price did not include any discounts or promotions.

3.1 Research Participants

This study recruited twenty adult internet users ages 20 or above, with online shopping experiences. Ten of them were from engineering programs, and the rest were from liberal arts programs. Those with experience in online auctions were given priority when recruiting. All participants were right-handed, and none took any medicine chronically. They filled out a personal internet experience questionnaire and signed the informed consent form before proceeding to the EEG experiment. Since the research design involved EEG, which demanded a sustained period of attention, the participants were required to avoid any activities that might disturb their biological indicators (e.g., having caffeinated drinks, taking medicine, or staying up late) for a 24-h period prior to starting the experiment. In addition, to better measure the participants’ brainwaves, they were instructed to wash their hair prior to the experiment to remove any hair products like hair spray or wax.

3.2 Research Procedures and Data Collection

The model of the caps procured for the experiment was the 40-channels Quik-Cap for NuAmps. The cap was connected to the NuAmps 40-channel Amplifier, an instrument that receives brainwave signals. The researcher used the software program E-Prime 2.0 to display the experimental materials on the screen in coordination with the EEG machine to collect behavior data, and the program automatically sent the data into Excel statistical software for analysis afterwards. During the experiment, the participants were randomly assigned to four experimental scenarios (one-two transaction products and with-without transaction labels). Each scenario contained 100 sets of images; each set contained three images: ISI (500 ms), target (1500 ms), ITI (1500 ms). The participants put on an electrode cap for capturing brainwaves (Fig. 2), signed the informed consent form, removed all metal materials, all hair accessories, and phones from their bodies. They were asked if they needed to use the restroom to avoid interruptions during the experiment. When putting on the cap, conductive adhesive was inserted between the electrodes and the scalp to aid signal reception. Then the EEG was used to measure the brainwave signal (Fig. 3). The research participants were instructed to observe the screen that showed the number of products available and the number of products desired. There was a break during the 1-h experiment. A compensation of $10 USD was given to participants who completed the entire process, and those who withdrew from the study midway received $3 USD to compensate for their time. After the experiment, the researcher used Excel to organize the behavior data and used SCAN 4.5 to analyze the brainwave data.

Fig. 2.
figure 2

The electrode cap used during the experiment.

Full size image
Fig. 3.
figure 3

EEG of Fz in simple scenario and scenario N.

Full size image

3.3 Scope and Limitations

This research study was designed based on the presumption that consumers (buyers) already had a purchase intention. It investigated how consumers, in the final steps of the purchase process, perceived different transaction mechanisms (product information texts or symbols designed for this study), and how these mechanisms affected the consumers’ decision-making efficiency, accuracy, and brainwave activity. The study adopted the EKB model and the final portion of the SOR model – consumers’ decision to buy – and excluded the other stages of the decision-making process. We also did not examine if the seller offered any promotions. In addition, the study controlled for similar online shopping contexts regardless of the actual auction platform (e.g., Yahoo, PChome, eBay, Taobao, Ruten.com, and Shopee). We also did not make a distinction of the product categories being considered, such as technology products, used versus new merchandize, etc. All participants had online purchase experiences. All participants were right-handed, and none took medicines chronically.

3.4 Research Data Analysis Methods

Orbital frontal lobe (Fz) and anterior cingular gyrus (FCz) are related to higher level cognition activities, thus they serve as the locations for the electrodes that detect brainwaves related to decisions. These areas are activated when making a purchase decision after considering purchase needs. However, when decision-making involves more uncertainties, the participants will be under pressure. The parietal lobe’s activities (CPz) process higher level perceptions, that is, they are sense-related activities, and the orbital frontal lobe cortex (Pz) and parietal lobe cortex (Cz) will be activated. Activities in the occipital lobe cortex and in the core of parietal lobe indicate a focused use of visual receptors.

First, the brainwave data were compiled and calibrated, and eye movements were removed. We then separated the EEG graphs into pre-stimulus (200 ms prior) and post-stimulus (1500 ms after) sections, using the pre-stimulus data as baseline for calibration. Noises (amplitudes that fell outside of the −100 μv–100 μv range) were removed. The brainwave data were filtered via band-pass filtering to keep the frequencies between 0.1 Hz and 40 Hz. Then, we analyzed the data collected using the Fz, FCz, Cz, Pz, and CPz electrodes, from the frontal lobe, to central sulcus, to the parietal lobe cortex which is responsible for visual space and working memory [32]. We also analyzed the average electric potential. Finally, focal analysis was conducted for the coding-phase P3 wave (the largest positive wave occurring between 300 ms and 450 ms after stimulus), the N2 wave (the largest negative wave occurring between 180 ms and 350 ms after stimulus), and the amplitudes and interpeak intervals for these two types of waves [11].

4 Findings

Table 1 organizes the average response time in different scenarios. One can see that regardless of the participants from engineering or liberal arts programs, there was a higher efficiency in terms of response time in simple scenarios (purchasing one item) than complex scenarios (purchasing two product items). However, in simple scenarios, both engineering and liberal arts students had more efficient response time for the scenario with textual transaction feedbacks (Scenario T) than without (Scenario N). While the engineering students had the shortest response time when dealing with the Scenario T, the difference between their response time and that of the liberal arts students was minimal. Similarly, in complex scenarios, both engineering students and liberal arts students had better response efficiency with the Scenario T than the Scenario N, but the engineering students had the longest response time with the Scenario N and the shortest response time with Scenario T.

Table 1. Average response time in various scenarios.
Full size table

Table 2 contains the results of analysis for various scenarios in terms of average response accuracy rate. With a 95% confidence interval, regardless of the scenario, the engineering students’ average response accuracy rate was greater than that of the liberal arts students, but students in either program had lower accuracy rate in complex scenarios than in simple scenarios. It is worthy to note that the engineering students had the best response time in the Scenario Simple and T; here, they also had the highest accuracy rate.

Table 2. Average accuracy rates of various scenarios.
Full size table

Figures 3, 4, 5, and 6 below show the results of the analysis on the EEG data, with the liberal arts students being represented by the red line in the graphs and the engineering students by the blue lines. The response time is the duration between the N2 wave (first negative wave) and the P3 wave (first positive wave). The Fz (cognitive) and Cz (perception) electrode data are presented to demonstrate the results of comparing between the different scenarios; the other electrodes’ graphs (FCz, Pz, CPz) are used for supporting evidence. In simple scenarios where one item was to be purchased, we compared the participants’ brainwaves in Scenarios N and T. We found that the liberal arts students’ Fz (cognitive decision-making) brainwave amplitudes were greater than those of the engineering students; their Cz (perception of stress) amplitudes were also greater that those of the engineering students; the former (cognition) was larger than the latter (perception).

Fig. 4.
figure 4

EEG of Fz in simple scenario and scenario T.

Full size image
Fig. 5.
figure 5

EEG of Cz in simple scenario and scenario N.

Full size image
Fig. 6.
figure 6

EEG of Cz in simple scenario and scenario T.

Full size image

In complex scenarios where two items were to be purchased (Figs. 7, 8, 9, and 10), we compared the participants’ brainwaves in Scenarios N and T. We found that the liberal arts students’ Fz (cognitive decision-making) as well as their Cz (perception of stress) brainwave amplitudes were again greater than those of the engineering students. Furthermore, the latter (perception) had greater amplitude than the former (cognition). This means that in complex scenarios, the liberal arts students’ felt text-related stress such that their brainwave amplitudes were more pronounced than their cognitive decision-making brainwave activities.

Fig. 7.
figure 7

EEG of Fz in complex scenarios and scenario N.

Full size image
Fig. 8.
figure 8

EEG of Fz in complex scenarios and scenario T.

Full size image
Fig. 9.
figure 9

EEG of Cz in complex scenario and scenario N.

Full size image
Fig. 10.
figure 10

EEG of Cz in complex scenario and scenario T.

Full size image

5 Discussion of Research Findings

Traditional views suggest that students in engineering fields can better process numbers, and students in liberal arts can better process texts. In the experiment, the findings show that regardless of the participants’ majors, they had more efficient performance (spending less time) and higher accuracy rates in simple scenarios than complex scenarios. Moreover, the engineering students’ average accuracy rates were greater than those of the liberal arts students. However, no visible difference in efficiency was found between the two groups of students. Specifically, both groups had shorter average response time when reacting to Scenario T than Scenario N. It can be inferred that having textual transaction feedbacks allowed participants to perceive more concrete information, thereby increased the response efficiency. Additionally, regarding the accuracy rates, we speculate that in general, engineering students possibly spend more time than liberal arts students in using computers for completing assignments and using labs, thereby demonstrated more composure when presented with either numbers or texts in the brainwave experiment.

The brainwaves change in the two groups of students showed that in simple scenarios, the liberal arts students had greater reaction to text-based stimuli than the engineering students in both cognitive decision-making and perception. Furthermore, the liberal arts students’ perceived stress might have affected their decision-making process. However, in complex scenarios, the liberal arts students’ brainwave amplitude was again greater than the engineering students’; in other words, in complex scenarios, the liberal arts students’ brainwave that perceive text-induced stress had greater amplitude than their brainwaves associated with decision and cognition. This indicates that, the difference between the liberal arts students’ and the engineering students’ perception of text-based stimuli was greater than the difference between the two groups’ cognitive decision-making brain activities. This is possibly because the liberal arts students’ perception of stress had more impact than their cognitive decision-making in a complex scenario.

Based on the findings of the study, we recommend the following future research and experimental procedures: (1) Individual differences: Based on the research participants and the research purpose, select appropriate cognitive assignments to better understand the differences among individual participants and treat the differences as variables in the experimental analyses. (2) Increase the sample size to reduce the effect of outliers and to draw more effective conclusions. (3) Attempt to increase the difficulty of the questions to increase the level of differentiation of participant response time, accuracy rate, and of the experiment as a whole. (4) Recruit participants from older age groups. This study involved mostly 20-year-olds, whose mental capacities are at a peak, making it harder to see differences in outcomes.

Finally, this experiment focused on the participants’ college major background, inferring that liberal arts students are more text-savvy and engineering students are more number-savvy. This is a traditional Taiwanese view: that students who choose liberal arts or social sciences as their majors generally are weaker in handling numbers and vice versa. However, the participants grew up in an era that emphasizes a well-rounded aptitude development and were all attending top colleges, hence they all had a high level of literacy and numeracy achievement. Simply using their college major categories as a variable was somewhat biased; therefore, it is recommended that future research that aims to differentiate participant characteristics to design the experiment differently to evaluate participant abilities.