Typically, people accept banknotes as change from another person or at a point-of-sale without consciously verifying authenticity (Klöne et al., 2019). Reasons for not checking authenticity are that counterfeit rates are extremely low, and that people trust the retailer (van der Horst et al., 2017a). Indeed, authentication may take place in a limited number of cases; for example, when the cash handler has encountered counterfeit banknotes before, or when the paper of a particular note feels somewhat unusual. Also, when one does not trust a particular transaction (e.g. an online purchase involving cash) one may check the authenticity of the banknote. A more practical constraint is that the general public has little knowledge of how to authenticate banknotes. On average, a person can mention two security features, but does not know what these features look like exactly, and where on a banknote these features may be found (van der Horst et al., 2017a). For instance, 69% of the general public knows that a euro banknote contains a watermark, but only 6% knows what image the watermark depicts (De Nederlandsche Bank, 2021a, 2021b). The next most known security feature is the hologram foil, mentioned by 39% of the public. The emerald number can be recalled only by 2% of participants.
As a consequence, a good deal of counterfeited banknotes goes undetected. To illustrate, van der Horst et al. (2017b) reported that around one in every five counterfeits is missed, in spite of the fact that participants were actively authenticating and were granted all the time they needed for this authentication task. It would not seem unreasonable to assume that the proportion of undetected counterfeits must be decidedly higher in everyday life, where cash handlers are not explicitly instructed to authenticate.
Yearly, the Eurosystem removes around 560 thousand counterfeits from circulation (out of a total of 24 billion banknotes; ECB annual report 2019). For an overview of the most prominent public security features, as indicated by the Nederlandsche Bank (DNB), see Fig. 1.
Yet another reason for not checking the authenticity of a banknote may be that the authentication process itself would constitute a socially awkward or uncomfortable situation—all the more fuelled by the fact that aforementioned lack of knowledge would likely make the authentication process a long one. If cash handlers were able to authenticate banknotes more quickly and covertly, it may well be that fewer counterfeits would go unnoticed. Additionally, if banknotes were authenticated more easily, perpetrators may be less inclined to use counterfeit banknotes in the first place.
In short, members of the public are rarely inclined to check a banknote for its authenticity, but when they do, they lack the capability to do it properly. Here we investigated whether counterfeit detection can be improved with the addition of novel, salient visual elements, designed to guide visuospatial attention to critical locations. Additionally, we assessed the impact of one’s a priori trust on attentional orienting.
Our hypotheses were guided by two distinct fields of study. The attention literature led us to reason that a counterfeited security feature should be detected more readily when attention is directed to the security feature’s location. One way to ensure that attention is directed to a critical location is to introduce a visually salient element near the location of the security feature such that attention is captured towards the critical location in a bottom-up way (e.g. Theeuwes, 2010; Wolfe et al., 2003). The hypothesized beneficial effect on counterfeit detection performance of having a salient element near a security feature, would be analogous to an attentional cueing effect (Posner, 1980). With respect to one’s a priori of trust, we reasoned that lower levels of trust would increase overall performance (due to increased effort). We were largely agnostic with respect to interactions between trust and cue validity. On the one hand, one might argue that increased effort (induced by low trust) would cause stronger attentional orienting and consequently stronger capture by salient design elements. On the other hand, an increased contribution of top-down attention might reduce the strength of bottom-up attentional capture. Let us now turn to these attentional dynamics.
Attentional processes in counterfeit detection
Cash transactions at a point-of-sale are generally performed quickly and automatically (van der Horst & Matthijsen, 2013). People do not give themselves time, or might feel embarrassed when scrutinizing the banknote (De Heij, 2017).
To authenticate a banknote properly, a good strategy is to direct attention to the security features. Attentional orienting can proceed in a bottom-up and top-down manner. Bottom-up attention is usually deployed reflexively due to the characteristics of the scene and stimulus saliency (e.g. Theeuwes et al., 2003), although the capture of attention can be prevented via an inhibitory mechanism that suppresses the salient stimulus (Luck et al., 2021). Top-down attention, which is thought to underly that inhibition, is usually deployed voluntarily in line with one’s tasks and goals (Egeth & Yantis, 1997). However, top-down authentication of banknotes is likely hampered by the handler’s aforementioned lack of knowledge.
It would therefore be ideal if security features were to capture attention in a rapid bottom-up manner (e.g. Theeuwes, 2019). It is worth noting that there has recently been a marked rise of simplified counterfeits without (mimicked) security features (Deutsche Bundesbank, 7–8-2020), suggesting that if attention were directed immediately and briefly to the relevant location on a banknote this could improve counterfeit detection. This underlines the importance of guiding banknote users’ attention to security features.
It may come as no surprise that saliency is a well-known concept among developers of banknote security features. For instance, nano-optic display technology features deliver a sense of movement, 3D depth, and multiple colours. According to manufacturers these technologies enable a wide array of custom design options to both capture and hold the user’s attention as they inspect and authenticate a banknote (16-11-2020, https://www.nanosecurity.ca/banknote-security/). However, to date there is no scientific dissemination about the effectiveness of security feature saliency. Furthermore, one must take into account the possibility that with increased saliency of one security feature, attention may increasingly be directed away from other security features. One challenge is thus to achieve optimally balanced saliency across features—a challenge enlarged by the fact that features differ from each other in terms of shape and size.
A potential solution—and the focus of this study—is to display a single type of salient element near each security feature. As such, the security features themselves can stay as they are, while the novel salient design element may become an established marker for areas worthy of inspection.
Although there is a lot of research suggesting that attention can be guided with the help of salient visual elements (e.g. Theeuwes, 2010), we must nonetheless be aware of one potential constraint. It is known that the most salient elements in a display typically receive attention first—irrespective of whether they are relevant or irrelevant (Wang & Theeuwes, 2020). Hence, if the salient element is at the same location as the security feature—as in the case of, say, a pink frame around the banknote’s emerald number—attention would be at the right location; but would it predominantly be directed to the pink frame, or to the emerald number itself? In the former scenario, the salient element would be helpful in roughly guiding attention (e.g. attention would be oriented to the right quadrant of the banknote), whilst interfering at a more detailed level (e.g. attention would be focused on the pink frame rather than on what is in the frame).
We chose the colour pink (desaturated red) for the frame, because of its saliency. In an experiment conducted by Drelie Gelasca et al. (2005) participants had to rank 12 colours in terms of saliency. The colours that had much more hits were red, yellow, green and pink. Those of lower saliency seemed to be light blue, maroon, violet and dark green. Also, in a colour experiment in which two groups searched for desaturated targets among saturated and white distractors, the conclusion was that the pink and peach targets have an advantage over the green, blue, and purple targets concerning reaction times (Kuzmova et al., 2008).
The impact of trust
As noted earlier, we expect that persons who have high trust in the authenticity of banknotes, for example because they assume that the counterfeit rate is low, perform worse than persons who expect a higher counterfeit rate. This hypothesis is based on the ‘prevalence-effect’. Observers tend to miss a disproportionate number of targets when these targets are rare (Wolfe & Van Wert, 2010). In everyday life, the prevalence of counterfeits is very low. The general public mentions this as an important reason for not authenticating (Klöne et al., 2019).
Lau and Huang (2010) found that the prevalence effect depends on past experience, not on future prospects. In their study, participants were told either that targets would be frequent (50%) or rare (10%), and both these instruction types were provided in settings where the true prevalence was either 50% or 10%; (hence, prevalence and the expectancy thereof were orthogonally manipulated). As it turned out, the error rate depended not on the instructions given but on the true target prevalence of the blocks. However, it might have been the case that participants simply did not believe the instructions (i.e. that expectancy was not successfully manipulated).
In fact, other research suggests that both target repetition and target expectancy play a role in the prevalence effect (Godwin et al., 2016). In the study of Godwin et al., one group of participants searched for low and high-prevalence targets of one particular colour throughout the experiment, while another group searched for one target colour on high-prevalence slides and a different target on low-prevalence slides. As such participants received differential levels of target repetition across the lower and higher-prevalence targets. An effect of prevalence emerged in both groups, although it was weaker in the single colour condition than it was in the alternating-colour condition, suggesting that both target repetition and target expectancy play a role in the prevalence effect.
Previous studies have shown that prevalence expectancy can simply be influenced by task instructions. For example, in their investigation of lesion detection on chest radiographs, Nocum et al. (2013) found that expectations of a higher abnormality-prevalence rate, as induced by instructions, impacted doctors’ perceptual sensitivity and visual search patterns, even though observers received the same stimulus material.
In the current study, we manipulated the expectancy of prevalence, which was assumed to affect top-down attention, and manipulated the presence or absence of a salient element around security features, which was assumed to affect bottom-up attention. The manipulation of expectancy is particularly important as it is one of the underlying factors of the trust one has in the payment system. The rationale is that people who have low trust in the authenticity of banknotes expect that the counterfeit rate is relatively high are more likely to invest more effort in authentication and thereby, to enhance authentication (van der Horst, et al., 2020a).
The present study
To summarize the above, typically the general public does not authenticate banknotes because they trust the banknote to be genuine and because they have insufficient explicit knowledge about which locations on the banknote inform its authenticity. Therefore, in this study, we examined whether salient elements around security features may help the public in authenticating a banknote at a quick glance. It is important to determine whether authenticating can be done rapidly because cash transactions typically occur within a very brief time frame (van der Horst et al., 2020b). We hypothesized that displaying a pink frame around a counterfeited security feature would lead to better counterfeit detection. This manipulation is to some extent analogous with the classic Posner exogenous cueing paradigm (Posner, 1980), in which targets are typically detected faster and more accurately when a cue is valid than when it is invalid.
Importantly, we did not instruct our participants on the existence and location of security features, as the general public is not trained either. Below it will be seen that overall detection scores were indeed not very high. However, our focus is not the performance per se, but the difference between having a salient element near to versus away from, the counterfeited feature, thought to operate as a valid versus invalid attentional cue, respectively. By directing the participants attention to a counterfeited feature, we expect to improve their ability to categorize the banknote as counterfeit.