Disentangling the P300 Components of Evoked Potentials Elicited using Subliminal Oddball Tasks in Reward-Directed Motivation

Kutlu Kaya; Ethem Gelir; Eda Karaismailoglu; Serkan Karaismailoglu

doi:10.4103/nsn.nsn_109_22

ORIGINAL ARTICLE

Year : 2022 | Volume : 39 | Issue : 4 | Page : 206-212

Disentangling the P300 Components of Evoked Potentials Elicited using Subliminal Oddball Tasks in Reward-Directed Motivation

Kutlu Kaya¹, Ethem Gelir², Eda Karaismailoglu³, Serkan Karaismailoglu¹
¹ Department of Physiology, Faculty of Medicine, Hacettepe University, Ankara, Turkey
² Department of Physiology, Faculty of Medicine, Ankara Medipol University, Ankara, Turkey
³ Department of Medical Informatics, Gulhane Faculty of Medicine, University of Health Sciences, Ankara, Turkey

Date of Submission	08-Jun-2022
Date of Decision	12-Oct-2022
Date of Acceptance	17-Oct-2022
Date of Web Publication	19-Dec-2022

Correspondence Address:
Kutlu Kaya
Department of Physiology, Hacettepe University Faculty of Medicine, 06230 Altindag, Ankara
Turkey

Source of Support: None, Conflict of Interest: None

DOI: 10.4103/nsn.nsn_109_22

Abstract

Objective: The objective of this study was to investigate neural responses during subliminal oddball tasks concerning reward-directed motivation to distinguish the P3a and P3b components of evoked P300 potentials. Methods: The subliminal oddball task included congruent/incongruent stimuli and masked prime subliminal stimuli. The task was to push the button when an incongruent stimulus appeared. Participants underwent two pre- and post-break electroencephalogram (EEG) recordings. During the break, the thirst scores of the participants were evaluated. Then, participants consumed three different salty foods in the same amount and completed the remaining two post-break EEG recordings, followed by the second thirst scores evaluation. Finally, participants physically selected lettered cards (A, B, C, and D) to receive a reward and quench their thirst. Results: Thirty participants were enrolled, 28 of whom were included for data analysis. Ten participants selected lettered cards as the given subliminal stimulus (sub⁽⁺⁾ group), and 18 participants selected different lettered cards from the given subliminal stimulus (sub^(–) group). We found a significant increase in post-P3a and post-P3b amplitudes in the sub⁽⁺⁾ group at the Pz/Oz electrodes. Changes in P3b amplitude were significantly higher in the sub⁽⁺⁾ group (2.83 ± 1.14 μV) than in the sub^(–) group (0.62 ± 2.29 μV) at the Pz/Oz electrodes. Correlation analysis revealed that higher thirst scores resulted in higher P3b amplitudes in the sub⁽⁺⁾ group. Conclusion: These findings suggest that reward-directed motivation increases parietal-posterior P3b amplitudes, signifying the involvement of cognitive processes to achieve a reward.

Keywords: Oddball, P300, P3a, P3b, reward, subliminal

How to cite this article:
Kaya K, Gelir E, Karaismailoglu E, Karaismailoglu S. Disentangling the P300 Components of Evoked Potentials Elicited using Subliminal Oddball Tasks in Reward-Directed Motivation. Neurol Sci Neurophysiol 2022;39:206-12

How to cite this URL:
Kaya K, Gelir E, Karaismailoglu E, Karaismailoglu S. Disentangling the P300 Components of Evoked Potentials Elicited using Subliminal Oddball Tasks in Reward-Directed Motivation. Neurol Sci Neurophysiol [serial online] 2022 [cited 2023 May 29];39:206-12. Available from: http://www.nsnjournal.org/text.asp?2022/39/4/206/364419

Introduction

Event-related potentials (ERPs) have been widely used for measuring cognitive responses to many repetitions of a given stimulus or condition.^[1] The well-known task used in ERP experiments is the oddball task, where participants are instructed to respond to target stimuli that randomly change among standard ones.^[2] The oddball task elicits P300 waves at midline electrodes (e.g., Fz, Cz, Pz, and Oz) and is assessed by changes in amplitudes (μV: microvolt) and latencies (ms: millisecond).^[3] Amplitude delineates the difference between the averaged baseline and the largest positive peak in a given time window, whereas latency defines the time between the stimulus onset and the largest peak. The amplitude and latency of ERPs provide important insight into cognitive functions because they can indicate the neural correlates of subliminal processing.^[4],[5] Subliminal stimuli are known to evoke early ERP components, including P100 and P200 waves.^[6] Nevertheless, the behavior of subliminal stimuli in the P300 components has not yet been disentangled.

Although the P300 wave alone reflects changes in working memory, there are two components of P300: P3a and P3b.^[7] The P3a elicits fronto-central (e.g., Fz and Cz) positive amplitudes and originates early (250–280 ms) in response to automatic attention and task demands, which is more likely to encompass subliminal information. Conversely, the P3b reflects parietal-posterior (e.g., Pz and Oz) positive amplitudes and originates later (300–400 ms), indicating more complex cognitive processes. Interaction between subliminal stimuli and P3a is expected, but subliminal P3b might be augmented via more complex tasks and individual reward-directed motivation.

There are three types of stimulus procedures in an oddball task: The single-stimulus procedure, the traditional two-stimulus oddball, and the three-stimulus oddball.^[8] In each procedure, the participant responds to the target stimulus and ignores the others. The three-stimulus procedure is similar to that of the classic oddball, but it is more challenging due to infrequently displayed compelling distracter stimulus in which the P3a and P3b decompose more noticeably. Moreover, maintaining attention to a goal provides more robust P300 waves and reliable subliminal findings. Therefore, reward-directed motivation increases the efficacy of subliminal stimuli.^[9],[10]

In this study, we employed a three-stimulus oddball task to distinguish P300 components of subliminal stimuli and behavior at P3a and P3b. We also urged our participants to complete the task to receive a reward. Thus, this study aimed to disentangle the relationship between reward-directed motivation (e.g., thirst-quenching) and P300 components.

Methods

Participants and procedure

Thirty volunteers participated in this study (mean age: 20.5 ± 0.58 years, range, 18–30 years; 12 females and 4 left-handed) through an advertisement on the university campus. Two participants were excluded due to excessive artifacts in the data. The Institutional Ethical Committee reviewed and approved this study (#GO 15/109-03). The study was explained to the participants by a researcher, and participants who met the inclusion criteria were given sufficient time to decide whether to participate and sign the informed consent. Study participants were asked to abstain from stimulant foods and beverages 2 h before the experiment. Participants were also informed at the beginning of the measurements that they were participating in a decision study where they would receive a beverage award upon completion regardless of their performance during the experimental procedure.

Electroencephalography (EEG) measurements were taken individually in a dimly lit, electromagnetically sound-attenuated Faraday cage between 11:00 AM and 2:00 PM. Participants were seated in a comfortable chair with a response button attached to the dominant hand, facing an LED display at a viewing distance of 120 cm. Before the EEG measurement, all instructions were explained to the participants, as shown in [Figure 1]a. The procedure included lowercase standard words (congruent stimuli), one uppercase word (incongruent stimulus), and a masked prime subliminal stimulus. The task was to push the button when the target appeared, an incongruent stimulus.

Figure 1: (a) Schematic representation of the study stimuli during the EEG recordings. The procedure included lowercase standard words (congruent stimuli), one uppercase target word (incongruent stimulus), and a masked prime subliminal stimulus. Each masked prime was preceded by a congruent word. Participants were instructed to push the button when the target appeared. (b) The experimental procedure started with two pre-break EEG recordings, followed by a break. Then, the thirst levels were surveyed, and the remaining two post-break EEG recordings were completed. Finally, participants selected lettered cards to receive a reward, which was a cold beverage

Click here to view

Each participant underwent two pre- and post-break EEG recordings. During the break, the thirst scores of the participants were evaluated using a 5-score scale: 1 = not thirsty, 5 = very thirsty. Subsequently, participants consumed three different salty foods, such as popcorn and chips, in the same amount. After post-break EEG recordings were completed, the thirst scores of the participants were evaluated a second time with the same survey. Finally, participants physically selected one lettered card (A, B, C, and D) to receive a beverage reward and quench their thirst. The experimental procedure is shown in [Figure 1]b. During post-processing, participants were divided into two groups: (1) participants who selected A or B cards with respect to the given subliminal stimulus (sub⁽⁺⁾ group, n = 10), and (2) participants who did not select the A or B cards with respect to the given subliminal stimulus (sub^(–) group, n = 18).

Stimuli

Stimuli were given using the Eevoke (Enschede, The Netherlands) software and were set in white color on a black background of 1920 × 1080 resolution with 120 font sizes in Arial Regular. The experimental stimuli consisted of nine words: seven standards, one target, and one masked prime. We conducted a total of four oddball tasks in Turkish. Masked prime subliminal stimuli were randomly displayed as either “seç A seç” (select A select) or “seç B seç” (select B select) to avoid bias. The target stimulus was “pencere” (window). Standard stimuli were “leylek” (stork), “erik” (plum), “gelenek” (tradition), “kutu” (box), “gemi” (ship), “ekmek” (bread), and “uzun” (long). The indication sequence of the stimuli was arranged randomly for each participant.

The duration of every visible stimulus was 200 ms and the interval was randomly changed between 1500 and 400 ms random interstimulus intervals to avoid anticipation. The fixed timing of 1500 ms was determined by studies in the literature, and the randomization time of 400 ms was constituted as twice our stimulus representation to decode the pattern of neuronal responses triggered by the subliminal presentation.^[11],[12] To achieve masked priming, nonsense letter strings were used before and after the prime stimulus for 71 ms (TsPLqA and WLuIMB, respectively).^[13] The masked prime was presented for 43 ms between nonsense letter strings [Figure 1]a. In each EEG recording, the masked prime was presented 10 times (a total of 80 times), standard stimuli were displayed 70 times (a total of 280 times), and the target stimulus was shown 10 times (a total of 80 times).

EEG recordings and analysis

The EEG sessions were recorded using 16-channel Ag/AgCl scalp electrodes embedded in an elastic cap with positions according to the 10–20 electrode system. Blinks and eye movements were monitored horizontally and vertically from three electrodes placed on the forehead (Fp1, Fpz, and Fp2). During the EEG recordings, electrodes with an impedance higher than 15 kOhm were discarded. The remaining electrodes (Fp1, Fpz, Fz, C3, Cz, C4, Pz, and Oz) were recorded with linked-ear reference and ground in the AFz electrode. The EEG was amplified using an ASA A.N.T. (Enschede, The Netherlands) and digitized at 1024 Hz. To process the EEG data, an in-house MATLAB (ver. R2018a, MathWorks, Inc.) pipeline was used for noise removal and averaging. Electrodes were re-referenced to the average of the recorded electrodes. The data were low pass filtered at 30 Hz using a second-order zero-phase shift Butterworth filter. Epochs that had large amplitudes of more than ± 100 μV were discarded. EEG waveforms were segmented in scanning regions from 0 to 500 ms in length. Then, an ERPLAB plugin was used for amplitude and latency quantification in EEGLAB.^[14],[15] The peak amplitude (μV) and the peak latency (ms) were measured for P3a in 250–280 ms and P3b in a 290–400 ms time window for each participant. Evoked potentials were then pooled at frontal/central sites (Fz and Cz electrodes) and at parietal/posterior sites (Pz and Oz electrodes) to calculate P3a and P3b responses.^[7]

Statistical analyses

All statistical analyses were performed using the IBM SPSS statistics package (ver. 23, IBM Corp, Armonk, NY, USA). The paired sample t-test was used to investigate changes in P3a and P3b amplitudes concerning pre- and post-break EEG recordings within each group. Using the independent t-test, we compared changes in P3a and P3b amplitudes between the sub⁽⁺⁾ and sub^(–) groups. Pearson's correlation test was used to reveal the relationship between the thirst scores and mean P3a-P3b values in each group. Statistical significance was evaluated using a two-tail t-distribution with a 95% confidence interval. To elucidate the relation between changes in thirst levels and P3a/P3b amplitudes in all pooled electrodes, a two-way repeated-measures analysis of variance (ANOVA) was performed using Greenhouse–Geisser correction. Data are reported as mean ± standard deviation unless stated otherwise, and P < 0.05 was considered significant.

Results

[Table 1] shows the changes in peak amplitudes in P3a and P3b at the Fz/Cz and Pz/Oz electrodes. [Figure 2] shows the grand averaged P300 measurement at the Pz electrode in the sub⁽+⁾ group. There were marked differences between pre- and post-P300 components. The amplitude of subliminal stimuli reached the peak value at 280 ms for pre-P3a and 320 ms for pre-P3b at the Pz electrode. Although P3a and P3b amplitudes were relatively the same during the pre-break, both amplitudes increased during the post-break, where amplitudes of P3b increased more than P3a [Figure 2]. The most noticeable difference was found at 330–340 ms, where the amplitude of post-P3b was higher than pre-P3b amplitude, indicating improved cognitive function.

Table 1: Peak amplitude changes in P3a and P3b at frontal/central (pooled Fz/Cz) and parietal/ posterior (pooled Pz/Oz) electrodes in each group

Click here to view

Figure 2: The grand averaged P300 signal in the sub⁽⁺⁾ group at the Pz electrode. Pre- and post-P300 signals are represented as a solid blue line and a dashed red line, respectively. The analysis period of P3a (250–280 ms) is represented in the blue-shaded area, and the period of P3b (>280–400 ms) is represented in the green-shaded area

Click here to view

At the Pz/Oz electrodes, post-P3a (5.62 ± 2.15 μV) and post-P3b amplitudes (6.79 ± 1.93 μV) were significantly higher than their pre-P3a (3.73 ± 1.76 μV, t(9) = −3.66, P = 0.005) and pre-P3b amplitudes (3.96 ± 1.86 μV, t(9) = −7.83, P < 0.001) in the sub⁽⁺⁾ group [Figure 3]. However, there were no significant changes in the sub^(–) group regarding pre- and post-P300 amplitudes. Between groups at the Pz/Oz electrodes, changes in P3b amplitude were significantly higher in the sub⁽⁺⁾ group (2.83 ± 1.14 μV) than in the sub^(–) group (0.62 ± 2.29 μV, t(26) = 2.83 P = 0.009). Nevertheless, there were no significant changes in P3a amplitudes at the Pz/Oz electrodes between the groups. At the Fz/Cz electrodes, there were no significant differences with respect to pre- and post-amplitudes. Latency analysis revealed no significant differences in all pooled electrodes (Fz/Cz and Pz/Oz) within and between groups concerning pre- and post-P300.

Figure 3: Peak amplitudes of pre- and post-P300 components (P3a and P3b) in each group at pooled Pz/Oz electrodes. Both post-P3a and post-P3b amplitudes significantly increased in the sub⁽⁺⁾ group compared with their pre-amplitudes. Post-P3b amplitudes in the sub⁽⁺⁾ group were significantly higher than post-P3b in the sub^(–) group (indicated as #, P = 0.009). Data are shown as mean ± standard error of the mean. *P < 0.05

Click here to view

We also assessed changes between pre- and post-break amplitudes and thirst scores in each group. In the sub⁽⁺⁾ group, the average post-thirst level (4.7 ± 0.48) was higher than that of pre-thirst (3.2 ± 0.79). At the Fz/Cz electrodes, two-way repeated-measures ANOVA with Greenhouse–Geisser correction determined that average post-P3b amplitudes (3.77 ± 2.15 μV) were significantly higher than pre-P3b amplitudes (3.52 ± 1.61 μV), accompanied by changes in thirst levels (F_1,9 = 7.67, P = 0.04). However, there was no statistical relationship between pre-P3a (3.32 ± 1.92 μV) or post-P3a (3.66 ± 2.14 μV) and changes in thirst (F_1,9 = 2.34, P = 0.16) in the same pooled electrodes.

At the Pz/Oz electrodes, we found the same results as the Fz/Cz electrodes with higher amplitudes for the sub⁽⁺⁾ group. We found that post-P3b amplitudes (6.80 ± 1.93 μV) were significantly higher than pre-P3b amplitudes (3.97 ± 1.86 μV) due to an increase in thirst scores (F_1,9 = 25.54, P = 0.001). However, there was no statistical relationship between pre-P3a (3.73 ± 1.76 μV) or post-P3a (5.63 ± 2.15 μV) and changes in thirst (F_1,9 = 0.41, P = 0.54) in the same pooled electrodes. Although there was a slight increase in post-thirst levels (3.3 ± 0.9) compared with pre-thirst (3.1 ± 0.8), there were no statistical differences between P3a or P3b amplitudes and changes in thirst level in any pooled electrodes for the sub^(–) group.

We found that changes in thirst scores were higher in the sub⁽⁺⁾ group (45% ± 48%) than in the sub^(–) group (14% ± 42%, t (17) = 1.70, P = 0.05). Pearson's correlation test was run to determine the relationship between thirst scores and amplitudes of P300 components. There was a moderate, positive correlation between thirst scores and P3b amplitudes in the sub⁽⁺⁾ group at the Pz/Oz electrodes, which was statistically significant (r = 0.55, P = 0.02) [Figure 4]. Although there was a moderate, positive correlation between thirst scores and P3a amplitudes in the sub⁽⁺⁾ group at the Pz/Oz electrodes, the relationship was not significant (r = 0.39, P = 0.09). No significant correlation was found between thirst scores and P300 amplitudes at the Fz/Cz electrodes in the sub⁽⁺⁾ group (r = 0.21, P = 0.1, and r = 0.24, P = 0.11, respectively).

Discussion

This study explores the distinct behaviors of P300 components in response to the subliminal oddball task and reward-directed motivation. Participants with higher thirst scores showed more attention allocation to the subliminal stimuli, as demonstrated by increased post-P3b amplitudes in conjunction with increased thirst levels. This study also demonstrated that (1) subliminal stimuli in an oddball task elicited P3a and P3b, and (2) how reward-directed motivation differentiated P3a and P3b.

Previous studies demonstrated the relation between masked priming and target stimuli in orthographic and phonologic domains,^[16],[17] semantic words,^[18] and numerical processing.^[13] Furthermore, a marked increase in the P300 amplitude of subliminal stimuli has been constantly shown in oddball tasks.^[13],[19] Furthermore, Strahan et al. demonstrated the relation between goal attainment and subliminal priming.^[9],[20]

There is straightforward relation between goal attainment and subliminal stimuli. However, the real beverage names or their derivatives have previously been used to assess the relation between subliminal priming and decision.^{[9],[21],[22]} Therefore, we supposed that displaying a beverage name or a nonsense word containing the same letters as the beverage would impact participants, resulting in a biased decision. Instead of employing a beverage name or its derivatives, we employed two letters separately (A and B) to be used as a subliminal stimulus, during which these letters were not visible in any part of our study. Hence, we hypothesized that the decision of a participant might be due to a simple subliminal message they were exposed to in relation to thirst levels. Thereby, we showed how to disentangle elicited P3a and P3b components.

P3a generates in the frontal-central brain areas, whereas P3b generates in the parietal-posterior brain areas, implying that the neural origins of P3a and P3b are different.^[23],[24] An oddball task requires discrimination between different stimuli (e.g., standard vs. target); therefore, it can elicit P3a effortlessly due to increased attention-driven neural activity, which requires frontal lobe activity.^[25] However, P3b occurs in more challenging tasks due to working memory engagement; a previous study found that P3b occurred in subliminal oddball tasks at midline electrodes and was closely related to attention and working memory.^[26] Similarly, we found more increase in P3b amplitude, suggesting a complex, maintained unconscious cognitive activity under subliminal processing facilitated by motivation. We also found that pre-P3a and pre-P3b were relatively the same. This was due to a basic attention allocation to target stimuli and masked prime stimuli without memory engagement. However, the higher post-P3a and post-P3b amplitudes strongly support that attention allocation to target stimuli is still present, but unconscious processes of masked prime are considerably increased.^[18]

Furthermore, the increase in P3b amplitude was higher in the sub⁽⁺⁾ group than in the sub^(–) group, signifying increased cognitive activity. However, the amplitude of post-P3a was relatively the same between the groups. This is due to subliminal stimuli entailing mainly attention processing, and activation in P3b indicates the participation of working memory.^[27] The late slow-wave (LSW) is another indicator of unconscious processing in P3b in the 600–1000 ms time window.^[28],[29] Although we did not analyze it in this study, we observed fast LSW after the P3b peak in the 400–600 ms time window. Because this is not solely a working memory study, this fast LSW may be because the memory workload is not increased. Nevertheless, we showed significant P3a- and P3b-driven subliminal processing in the sub⁽⁺⁾ group in response to reward-directed motivation, which was not observed in the other group.

Participants were notified that there was a reward after tasks were completed, which was a cold beverage. This was executed to break the participants' anticipation and engaged them toward a goal. We evaluated the relation between subliminal stimuli and reward engagement with two aspects. First, we demonstrated that amplitudes of post-P3b both in Fz/Cz and Pz/Oz pooled electrodes were higher than in the pre-P3b in the sub⁽⁺⁾ group, accompanying an increase in thirst scores. However, we did not observe the same relation between pre- and post-P3a amplitudes in the sub⁽⁺⁾ group. Moreover, there was no statistically significant relation between pre- and post-P3a/P3b amplitudes in any pooled electrodes in the sub^(–) group. These results imply that there is increased attention and improved cognitive involvement in subliminal stimuli, triggered by thirst and reward engagement. Second, we performed a correlation analysis to uncover the relationship between thirst scores and P300 amplitudes. The correlation analysis revealed a significant linear relationship at Pz/Oz electrodes. We found that the higher thirst scores resulted in increased P3b amplitudes in the sub⁽⁺⁾ group. The presence of metabolic needs (e.g., thirst-quenching) enhances subliminal processing.^[9],[21] Because the thirst scores were higher in the sub⁽⁺⁾ group, this suggests that satisfying a metabolic need caused more allocated attention and working memory activation. Moreover, the exact relationship was not seen in the sub^(–) group. This observation confirmed that subliminal processing was altered by underlying physiologic needs, which are important and form cognitive processes.

The results from this study are limited to the neural responses from the midline electrodes. ERPs from more electrodes would be interesting to compare the hemispheric responses and help identify the lateralization of each response. We also did not aim to increase the cognitive workload, which might help discriminate more P300 components and the differences between stimuli. Future research should focus on greater EEG resolution and neurovascular responses, incorporating temporal and spatial variations in neurovascular coupling during subliminal oddball tasks. Further investigation is needed to elucidate neurovascular coupling in subliminal processing using advanced algorithms, such as machine learning.

Conclusion

The results of this study illustrate the behavior of P300 components, P3 and P3b. We have shown that subliminal encoding can reliably reflect underlying neuronal activity during oddball tasks. Moreover, marked changes in P3a and P3b can be differentiated by reward-directed motivation. This, therefore, leads us to conclude that reward-directed motivation that promotes subliminal stimulus recall increases the amplitude of P300 components. Overall, as illustrated in this study, the differences between the groups under the same conditions provide new insight into subliminal studies.

Acknowledgment

K.K., E.G., and S.K. designed the study. K.K. and E.G. conducted the study. K.K. processed and analyzed the EEG-ERP data. E.G. and S.K. helped interpret the analyzed data. K.K. and E.K. ran the statistical analyses. K.K. wrote the initial manuscript with input from E.G and S.K. All authors provided intellectual inputs to the manuscript and reviewed and read the final manuscript.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

References

1.	Sur S, Sinha VK. Event-related potential: An overview. Ind Psychiatry J 2009;18:70-3. [PUBMED] [Full text]
2.	Ebmeier KP, Steele JD, MacKenzie DM, O'Carroll RE, Kydd RR, Glabus MF, et al. Cognitive brain potentials and regional cerebral blood flow equivalents during two- and three-sound auditory “oddball tasks”. Electroencephalogr Clin Neurophysiol 1995;95:434-43.
3.	Johnson R Jr. On the neural generators of the P300 component of the event-related potential. Psychophysiology 1993;30:90-7.
4.	Bernat E, Shevrin H, Snodgrass M. Subliminal visual oddball stimuli evoke a P300 component. Clin Neurophysiol 2001;112:159-71.
5.	Shevrin H. Event-related markers of unconscious processes. Int J Psychophysiol 2001;42:209-18.
6.	Lamy D, Salti M, Bar-Haim Y. Neural correlates of subjective awareness and unconscious processing: An ERP study. J Cogn Neurosci 2009;21:1435-46.
7.	Polich J. Updating P300: An integrative theory of P3a and P3b. Clin Neurophysiol 2007;118:2128-48.
8.	Polich J, Criado JR. Neuropsychology and neuropharmacology of P3a and P3b. Int J Psychophysiol 2006;60:172-85.
9.	Strahan EJ, Spencer SJ, Zanna MP. Subliminal priming and persuasion: striking while the iron is hot. J Exp Soc Psychol 2002;38:556-68.
10.	Aarts H, Dijksterhuis A, De Vries P. On the psychology of drinking: Being thirsty and perceptually ready. Br J Psychol 2001;92:631-42.
11.	Teixeira M, Pires G, Raimundo M, Nascimento S, Almeida V, Castelo-Branco M. Robust single trial identification of conscious percepts triggered by sensory events of variable saliency. PLoS One 2014;9:e86201.
12.	Sperdin HF, Spierer L, Becker R, Michel CM, Landis T. Submillisecond unmasked subliminal visual stimuli evoke electrical brain responses. Hum Brain Mapp 2015;36:1470-83.
13.	Naccache L, Dehaene S. The priming method: imaging unconscious repetition priming reveals an abstract representation of number in the parietal lobes. Cereb Cortex 2001;11:966-74.
14.	Lopez-Calderon J, Luck SJ. ERPLAB: An open-source toolbox for the analysis of event-related potentials. Front Hum Neurosci 2014;8:213.
15.	Delorme A, Makeig S. EEGLAB: An open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods 2004;134:9-21.
16.	Bijeljac-Babic R, Biardeau A, Grainger J. Masked orthographic priming in bilingual word recognition. Mem Cognit 1997;25:447-57.
17.	Ferrand L, Humphreys GW, Segui J. Masked repetition and phonological priming in picture naming. Percept Psychophys 1998;60:263-74.
18.	Dehaene S, Naccache L, Le Clec'H G, Koechlin E, Mueller M, Dehaene-Lambertz G, et al. Imaging unconscious semantic priming. Nature 1998;395:597-600.
19.	Devrim M, Demiralp T, Kurt A. The effects of subthreshold visual stimulation on P300 response. Neuroreport 1997;8:3113-7.
20.	Strahan EJ, Spencer SJ, Zanna MP. Subliminal priming and persuasion: how motivation affects the activation of goals and the persuasiveness of messages. Applying Social Cognition to Consumer-Focused Strategy (1^st Ed). Psychology Press, New York; 2005. p. 267-80.
21.	Karremans JC, Stroebe W, Claus J. Beyond Vicary's fantasies: The impact of subliminal priming and brand choice. J Exp Soc Psychol 2006;42:792-8.
22.	Bermeitinger C, Goelz R, Johr N, Neumann M, Ecker UK, Doerr R. The hidden persuaders break into the tired brain. J Exp Soc Psychol 2009;45:320-6.
23.	Soltani M, Knight RT. Neural origins of the P300. Crit Rev Neurobiol 2000;14:199-224.
24.	Polich J, Comerchero MD. P3a from visual stimuli: Typicality, task, and topography. Brain Topogr 2003;15:141-52.
25.	Curran T. Effects of attention and confidence on the hypothesized ERP correlates of recollection and familiarity. Neuropsychologia 2004;42:1088-106.
26.	Silverstein BH, Snodgrass M, Shevrin H, Kushwaha R. P3b, consciousness, and complex unconscious processing. Cortex 2015;73:216-27.
27.	Kok A. On the utility of P3 amplitude as a measure of processing capacity. Psychophysiology 2001;38:557-77.
28.	Gable PA, Adams DL. Nonaffective motivation modulates the sustained LPP (1,000-2,000 ms). Psychophysiology 2013;50:1251-4.
29.	Strüber D, Polich J. P300 and slow wave from oddball and single-stimulus visual tasks: Inter-stimulus interval effects. Int J Psychophysiol 2002;45:187-96.