Memory is one of the fundamental elements of creativity (Beaty et al., 2017; Benedek and Fink, 2019). Creativity cannot come ex nihilo; instead, it is a process where novel ideas are generated by searching (Fink and Benedek, 2014), interacting (Palmer, 2020) and associating (Benedek and Fink, 2019) existing memories. In some creative thinking processes, before generating new ideas, designers may consciously search the information in their mind to extract useful information. As the source of new ideas, memory has been identified with the activity of amygdale by functional magnetic resonance imaging (fMRI) (Dinar et al., 2015). Alpha (Ali et al., 2022), theta (Wang et al., 2022), and gamma waves (Sharpe and Mahmud, 2020) were detected to be related to the memory processes. It is important to note that memory processes can be divided into two types. One process is encoding memories, and the other is extracting useful memories. In this study, the process referred to is the extracting memory process rather than the encoding process. funding Memories can be further divided into long-term memory (LTM), short-term memory (STM), and sense memory. Sense memory is the information that is acquired through hearing, vision, touch, and other senses (Di Benedetto, 2007). Few research studies have focused on identifying the relations between sense memory and creativity because sense memory is an unpredictable process and is hard to be controlled in labs.

Short-term memory has a longer processing time (a few seconds or a few minutes) than sense memory. Unlike sense memory, which works unconsciously, STM allows people to repeatedly and consciously recall facts and events (Norris, 2017). STM is involved in creative processes because creativity processes require to store information temporarily (Mao et al., 2020). Gubbels et al. (2017) identified the relationship between STM, analytical ability, and creativity. They asked children to look at 20 graphics and remember them. Then, participants were asked to write down a description of each graphic based on their memory. In another task, participants were asked to write down 10 words they heard and then rewrite them based on what they could remember. The results showed that (visual and language) STM affects the analytical ability and the level of creativity. The reason why it is hard to test STM ability in a creativity task is that participants were not able to report their STM processes in a creative task as the reporting process will interrupt the STM process. This is the reason why neuroscience technologies are needed for reporting tasks. With the help of neuroscience methods, it is found that STM is positively associated with delta and theta waves and negatively associated with alpha waves (Trammell et al., 2017).

Long-term memory is the memory that has been stored in the brain for a long time (Norris, 2017). LTM has been associated with creativity because LTM includes information about previous knowledge. This knowledge can be used to create ideas that are related to creativity tasks (Goldschmidt, 1995). LTM can be divided into declarative memory and non-declarative memory. Researchers have found that creativity is related with LTM, especially declarative memory (Benedek et al., 2020). Declarative memory is the memory that people can access consciously. It can be further divided into semantic memory and episodic memory. Semantic memory is the memory of facts which will not be changed or limited by time and space. For example, the current capital of China is Beijing. Semantic memory is considered to have a relationship with creativity for it can provide information about facts and concepts, acting as a source of creativity to generate new ideas (Kenett and Faust, 2019). Semantic memory can also support the association of related concepts (Volle, 2018). The association process may also be helpful to generate creative ideas (Huang et al., 2015; Beaty et al., 2020).

Episodic memory is the memory that an individual experiences in a specific time and location, such as information on daily life; for example, the memory “today, I drank a glass milk in the morning.” Episodic memory can support the generation of creative ideas (Madore et al., 2016) because it allows people to stimulate their memories of previous events (Benedek and Fink, 2019) or provide details of previous events as the creativity stimulation (Madore et al., 2015). Moreover, in the episodic memory process, people do not simply search their memory and select useful sources individually. Episodic memory processes also restructure memories in the brain. In other words, the event that people recall is not the original event that people went through; instead, it has been processed by the brain’s episodic memory process. The retrieval and reconstruction processes act as a source of creativity and helps to stimulate the imagination (Beaty et al., 2020).

Researchers have investigated the application of neurotechnology to identify the relationship between semantic or episodic memory and creativity, some of whom have focused on the neural structure of the brain. When engaged in the creative process, the default mode network (DMN), which is related to semantic and episodic memory, will be activated (Benedek and Fink, 2019). With the help of fMRI, semantic memory in a creative process was further identified to be related to the left angular gyrus, left inferior parietal lobule and posterior cingulate cortex, while episodic memory in a creative process was further identified to be related to left parahippocampal gyrus and right inferior parietal lobule (Beaty et al., 2020).

For convenience, the following terms in this study are used: “retrieval” is used to represent the episodic memory process, as episodic memory is related to the retrieval of a previous event; “recall” is used to represent semantic memory process, as semantic memory is related to recalling the knowledge of a fact.

Many researchers have realized the importance role of the cognitive process of association in creativity (Guilford, 1956; Finke et al., 1992; Wu et al., 2021; Yin et al., 2022). Association comes in the forms of remote and common associations (Benedek et al., 2020). Remote association is the ability to associate unrelated concepts while common association is the ability to associate related concepts. Remote association has a positive effect on creativity (Liu, 2016; Benedek et al., 2020). The activity of alpha waves was identified to be related with remote association (Fink et al., 2009). Researchers have also tried to identify which parts of the brain are related with remote association. However, studies on the involvement of remote association have not found consistent results. Fink et al. (2009) suggested that the left frontal lobe was active when people make use of remote association in creative processes (Purcell and Gero, 1998). Stevens and Zabelina (2019) supported the view that the left temporal lobe is related to remote association processes in creativity, whereas Jung-Beeman (2005) thought it was the right temporal lobe that relates to remote association processes. Common association can also contribute to the creative ideas. The differences between remote association and common association are on the creative ideas quantity (Purcell and Gero, 1998). Fewer creative responses were generated through common association processes compared to remote association processes. In addition, compared with remote association, the activation levels of brain and the alpha waves during common association process is weak (Purcell and Gero, 1998; Stevens and Zabelina, 2020).

Combination is also considered to be one of the cognitive processes involved in creativity. Combination ability is related to attention and LTM. When people have broad attention, they have a better chance to combine relative information with new concepts (Carson et al., 2005). Additionally, combining concepts is one of the operations controlled by LTM (Simon and Simon, 1978), especially with regard to episodic memory (Kenett and Faust, 2019), because they are the source from which a combination can lead to a concept. Researchers have applied quantitative studies to identify the relations between combination process and creativity (Wan and Chiu, 2002). Results have shown that the creativity score of novel combination tasks is higher than that of common combination tasks. However, few studies so far have applied neuroscientific methods to identify the relationship between combination and creativity.

Different neurophysiological characteristics of creativity-related cognitive factors (recall, retrieval, combination, and association) have been identified. However, the existing studies mainly identified the relationship between cognitive factors and creativity singly, focusing on which part of the brain is activated in the process and which type of wavebands are related. Other neurophysiological characteristics, such as event-related potentials (ERPs), have not been fully studied. Also, the neurophysiological characteristics of creativity-related combination processing have not been studied in detail. Therefore, this study aims to understand the EEG characteristics of creativity-related cognitive factors (recall, retrieval, combination, and association) in this way to better understanding the creativity process from neurophysiological levels.

In total, 30 right-handed Chinese participants (15 female, 15 male, aged 20–25), experienced in industrial or product design, were recruited. Their creativity levels were reported from the Epstein Creativity Competencies Inventory for Individuals (ECCI-i), which is a 28-items 5-point Likert-type scale labeled from Strongly Disagree (1 point), Disagree (2 point), Neither (2 point), Agree (4 point) to Strongly Agree (5 point; Epstein et al., 2008). All participants’ ECCI-i scores were over 120, which indicated that the participants have a strong potential to have creative ability (Epstein, 2000).

Before the study, all participants were ensured through self-reporting to have normal or corrected-to-normal vision and have no diagnosed psychiatric disorders, color blindness or other barriers to using computers. Also, it was ensured that the participants did not ingest any caffeine, unprescribed medication or alcohol in the previous day before taking part in this study. After the study, all participants self-reported that they had not seen the design tasks finished in this study before and had expressed their ideas clearly. This study was approved by the local ethics committee of the first author (reference number: 20IC6227).

The study included five induction tasks to test participants’ remote association, common association, combination, retrieval, and recall ability. The procedure of the study is shown in Figure 1.

A medical-grade EEG device, the Neurofax EEG-9200 system, was used to record the EEG signals (NIHON KOHDEN, Tokyo, Japan). The Neurofax EEG-9200 system includes 16 scalp and 2 mastoid Ag/AgCl electrodes mounted according to the 10/20 system (Figure 2). Also, it includes an EEG measurement system, an amplifier and an EEG result viewing software. The impedances of all the EEG channels were below 5 kΩ. The data were sampled at 1,000 Hz. The EEG tasks were generated and presented with the help of E-Prime 3.0. All tasks were presented on a computer screen (35.89 × 24.71 cm with a resolution of 2,560 × 1,600). The data were collected and stored in the Neurofax EEG-9200 system.

Because the EEG electrodes are relatively far from the neurons where the signals are generated, EEG has a relatively low spatial resolution (Srinivasan, 1999). Therefore, increasing the number of electrodes may only provide diminishing returns in terms of EEG data acquisition. In other words, with more electrodes, the correlated signals may tend to interfere with adjacent channels. Statistical methods are often used to combine signals of interest coming from multiple channels into a single signal. In other words, increasing the spatial sampling density (channel quantity) means more channels will be included in that cluster, instead of increasing the efficiency of the data. Therefore, more channels may not generate effective EEG data. A further issue with this is that more channels increase the processing time for the data that are stored and analyzed.

According to previous research, the areas of the brain that relate to creativity-related cognitive processes (remote association, common association, retrieval, combination, and recall) have been found. The results are visualized in Figure 2. Remote association was related to the right temporal lobe (Jung-Beeman, 2005), left frontal lobe (Purcell and Gero, 1998; Fink et al., 2009), and occipital cortex (Boccia et al., 2015). Common association was related to the inferior parietal lobe (Benedek et al., 2014). Retrieval was related to the medial temporal lobe (Madore et al., 2016; Beaty et al., 2020). Recall was related to the frontopolar cortex (Green, 2016; Beaty et al., 2020). There was no research to identify which areas of the brain were stimulated by combinations of these cognitive processes.

Compared with the finding areas and the areas covered by the 16 EEG channels, most of the identified areas were covered. Thus, more channels may not be necessary. To be specific, channels Fp1/Fp2/F7/F8/F3/F4 report signals on the frontal lobe, C3/C4/P3/P4 report signals on the parietal lobe, T3/T4/T5/T6 report signals on the temporal lobe and O1/O2 report signals on the occipital lobe. Considering that the study is also interested in potential hemispheric differences, midline electrodes such as FZ, CZ, and PZ, were not included (Schwab et al., 2014).

The MATLAB R2018b (The MathWorks Inc., Natick, MA, United States) plugin EEGLAB was used to analyze the signals. A 50 Hz notch filter was applied to negate the interference of the electrical mains. Then, the signals were passed through a band-pass filter with a pass-band of 0.1–100 Hz (Zarjam et al., 2011; Schwab et al., 2014). The reference electrodes were placed on the left and right mastoid processes.

The study then compared the ERP and active brain areas of the five EEG events (remote association, common association, combination, retrieval, and recall) by marking and extracting them from the EEG signals. The ERP subject averages for each event were the averaged results of all participants and all event-related task trials.

Spectral analysis was conducted for each event. The percent relative variance, sometimes called the relative variance, was calculated. The spectral results of each event are shown in Figure 3. Some previous studies have suggested that percent relative variance can be used to define the effect of a particular variable on the whole condition (Hermance, 2013). In EEG, the component percent relative variance was used to represent the contribution of a specific component on a particular channel or the whole channels (Delorme and Makeig, 2004). Therefore, in this study, the percent relative variance defines the effect that a particular component of EEG on the whole EEG channel or a particular individual channel.

The component X (1 - 16) from the spectral analysis is the same component from the independent component analysis (ICA) results. The ICA results of each event are shown in Figure 4. Since the percent relative variance reported the effect that a particular component of EEG on the EEG channels, to identify which brain areas were related to a specific cognitive-factor event, in each specific cognitive-factor event, the component percent variances were ranked. Then, the highest component percent variance that related to specific brain areas was used as the cue to identify which brain areas were activated in the specific cognitive-factor event.

The results demonstrate that the cognitive processes of remote association, common association, recall and retrieval mainly relate to the frontal lobe (Fp1 and Fp2 channels; based on component 5, component 8, component 5, and component 8, respectively). The combination process mainly related to the left frontal lobe (Fp1 channel; based on component 1).

The ERPS results of each event were analyzed based on the related activated brain area EEG channels from the spectral results. Specifically, remote association, common association, retrieval and recall events are based on the Fp1 and Fp2 channels. The combination events are based on the Fp1 channel.

The results have been presented in Figure 5. The highest ERP for the remote association event was generated at 164 ms. The highest ERP for the ERP was generated at 95 ms. The highest ERP for the combination event was generated at 1,293 ms. The highest ERP for the retrieval event was generated at 2,320 ms. The highest ERP for the recall event was generated at 311 ms.

Although the location information of EEG is not that accurate, it does, to some extent, reflect the active brain areas and thus is compared to existing research.

Event-related potential can quantitatively reflect the brain’s response to a specific cognitive event (Sa, 2005). From the ERP results, a few findings can be summarized.

This study has some limitations. Firstly, this study only recruited 30 Chinese participants. The participants’ culture and ages may have affected the EEG results. Therefore, in the future, more participants from different age groups and cultures should be incorporated.

Secondly, there was an attempt to conduct the study without any external interference (such as motion and noise). However, the study cannot rule out the possibility of spill-over effects completely. In other words, it may be possible that the previous task/trial may have affected later ones. What could be done in future studies is to limit the spill-over effects by presenting the inducted tasks in a random order and presenting the trials in each task in a random order.

Thirdly, the study followed the guided assumption from Beaty et al. (2020) that participants could follow the instructions completely. Also, the study assumed that the EEG recorded in each cognitive factor task represented participants’ cognitive factor ability completely accurately. In other words, the study hypothesized that participants did not thinking of anything that was not related to the cognitive factor tasks and that their thinking of process relied on their cognitive factor-related ability. However, it is hard for researchers to objectively check whether participants did not have thoughts unrelated to the cognitive factor tasks or whether their thought processes solely relied on their cognitive factor-related abilities. Therefore, whether the identified EEG signals solely represented the actual cognitive factor-related EEG signals cannot be ensured. This makes the results less reliable. Even if the study assessed the cognitive factor task results, it would only reflect the participants’ creativity levels; the researchers would still be unsure whether the results were generated from the cognitive factor task-related abilities. Therefore, future studies should add a checking mechanism to increase the accuracy of the EEG quality. For example, after each trail or task, an interface can be displayed and asked participants whether they thought anything that was not related to the cognitive factor tasks and whether they thought relied on their cognitive factor-related ability completely.

Moreover, this research included the ideas (or concepts), which generated from the neural activity in induction processing, would be identical to those from the creative design process, especially for the recall and retrieval induction tasks. Whether the actual condition is the same as what the research hypothesized needs to be further detected. To reduce the bias generated from this limitation, this study adapted the induction tasks that have been done by existing research. Also, all of the selected induction tasks were mentioned to have the ability to identify the relations between a specific cognitive factor and creativity. However, whether the selected induction tasks have this ability was not completed studies and the limitations may still exist.

Finally, the EEG results were collected from a medical-grade, 16 channels EEG device. This EEG device is used for in the industry for medical diagnoses. Therefore, the signal quality is different to EEG devices used in other creativity studies by universities, corporate laboratories and national research institutes, which using the non-clinical-level EEG devices. This higher quality signal can mitigate the limitation of having fewer channels, to some degree. In addition, although from the previous explanation of the diminishing returns of multiple channels, these 16 channels were considered to be enough for the study. However, the researchers could not know if the most effective number of channels is 16 or not. In other words, although the collected results have a tendency toward saturation with a number increasing of channels, whether the saturation point is at 16 channels is not clear. Therefore, studies in the future should experiment with various numbers of channels.