TrunkpersName, StefanicspersName, ZentaipersName, BacskaypersName, FelingerpersName, ThuróczypersName, HernádipersName (2015date). “Effects of concurrent caffeine and mobile phone exposure on local target probability processing in the human brain.” Scientific Reportstitle·j.
Object of study — whether caffeineterm and UMTSterm mobile-phone irradiation interact to affect target probabilityterm·var processing
Research question — whether combined caffeineterm and mobile phoneterm·key exposure affects reaction timeterm and pre-targetterm·key brain oscillations more than either alone
Contribution — a full-factorial, pharmacologically validated null on mobile-phone exposure, alongside positive caffeineterm effects on reaction timeterm and alphaterm amplitude
phenomenon · claim-type · cross-domain-transfer · mechanism · claim-relation · replication-status · limitation — grounded in the full text| phenomenon | caffeine and mobile-phone RF exposure effects on visual target detection — “Here we investigated whether caffeine intake and concurrent exposure to Universal Mobile Telecommunications System (UMTS) MP-like irradiation may interactively influence neuro-cognitive function in an active visual oddball paradigm.” | |
| claim-type | null-result — “However, in the present pharmacologically validated study UMTS exposure either alone or in combination with caffeine did not alter RT or pre-stimulus oscillatory brain activity.” | |
| method — evidence | — “In a full factorial experimental design, 25 participants performed a simple visual target detection task while reaction time (RT) and electroencephalogram (EEG) was recorded.” | |
| object of study — evidence | — “Millions of people use mobile phones (MP) while drinking coffee or other caffeine containing beverages. Little is known about the potential combined effects of MP irradiation and caffeine on cognitive functions.” | |
| research question — evidence | — “Our hypothesis was that, due to previously reported 1 , 6 similar facilitatory effects on brain excitatory activity, simultaneous caffeine and MP exposure will have a larger effect than caffeine or MP EMF exposure alone.” | |
| contribution — evidence | — “However, in the present pharmacologically validated study UMTS exposure either alone or in combination with caffeine did not alter RT or pre-stimulus oscillatory brain activity.” | |
| transfer | from pharmacology: the additive-model test for drug synergy, applied to a drug and a physical exposure — “To reveal potential interactions of caffeine and UMTS MP exposure in RT and oscillatory measures we focused on the exposure block and we used the additive analysis model with the following formula, as applied elsewhere 42 , 43 , 44 : Hereafter, “[Caffeine − Control] + [UMTS − Control]” and “Combined − Control” are referred to as ‘sum’ and ‘simultaneous’ data, respectively. We hypothesized that violation of the additivity of the analyzed RT or spectral amplitude measures would indicate synergistic interactions.” | transfers[0] |
| transfer | from analytical chemistry: HPLC assay of saliva, used to verify that the pharmacological manipulation actually took effect — “Saliva samples were taken at the beginning and the end of each recording session and caffeine concentrations were determined by high-performance liquid chromatography (HPLC).” | transfers[1] |
| mechanism | Caffeine raises the arousal state, which suppresses pre-target alpha oscillations and thereby improves visual perceptual performance — “One possible explanation of the decreased alpha band activity in the pre-target by caffeine may be that caffeine increased the arousal state, thus promoted better visual perception performance compared to the control condition 54 .” | mechanisms[0] |
| mechanism | Caffeine increases sensitivity to the regularity of the stimulus sequence, making implicit short-term memory for the preceding standards more efficient and so speeding responses to expected targets — “Thus, it would be reasonable to conclude that caffeine increases the sensitivity to the stimulus pattern, and improves the efficiency of implicit short-term memory 41 .” | mechanisms[1] |
| mechanism | Alternative account: caffeine acts only at the motor end, speeding the initiation of an already prepared response rather than altering expectancy-related cognition — “Namely, caffeine only facilitates the faster initiation of an already prepared response, independently of the underlying differential cognitive processes (e.g., higher or lower stimulus expectancy).” | mechanisms[2] |
| mechanism | A drop in gamma-band activity before an anticipated target frees processing for the incoming task-relevant stimulus, speeding its detection — “Thus, it is likely that the decreased gamma activity prior to the predicted arrival of a target stimulus may facilitate the processing of relevant task-related information 36 .” | mechanisms[3] |
| mechanism | Caffeine and mobile-phone EMF were expected to combine because both act on calcium-dependent signalling, which underlies cognitive function — “While calcium plays an important role in cognitive functions 20 , 21 , 22 , any combined effects of caffeine and MP exposure on calcium related mechanisms may affect cognitive performance indexed by reaction time and brain oscillatory activity.” | mechanisms[4] |
| mechanism | Proposed reason for the UMTS null result: the modulation used was biologically ineffective or the field intensity fell below the threshold for any detectable biological effect — “A possible explanation for lack of UMTS exposure effects may be that the applied signal modulation was ineffective or the signal intensity was too low, i.e., under the threshold for detectable biological effects.” | mechanisms[5] |
| mechanism | caffeine is used as a pharmacological positive control, on the grounds that this dose reliably improves cognition — “Caffeine exposure also served as pharmacological (positive) control as it has been previously reported that intraoral administration of 3 mg/kg b.w. caffeine reliably improved cognitive functions and mood” | mechanisms[6] |
| mechanism | saliva caffeine concentration is measured by HPLC at the start and end of each session, so the exposure is verified rather than assumed — “Saliva samples were taken at the beginning and the end of each recording session and caffeine concentrations were determined by high-performance liquid chromatography (HPLC).” | mechanisms[7] |
| mechanism | an additive model tests synergy directly: the sum of the two single-agent differences is compared against the combined condition — “To reveal potential interactions of caffeine and UMTS MP exposure in RT and oscillatory measures we focused on the exposure block and we used the additive analysis model with the following formula, as applied elsewhere 42 , 43 , 44 : Hereafter, “[Caffeine − Control] + [UMTS − Control]” and “Combined − Control” are referred to as ‘sum’ and ‘simultaneous’ data, respectively. We hypothesized that violation of the additivity of the analyzed RT or spectral amplitude measures would indicate synergistic interactions.” | mechanisms[8] |
| mechanism | dose is individualised at 3 mg/kg body weight, averaging 211.56 mg — “Three mg/kg caffeine packed in identical hard gelatin capsules were administered to the participants. The capsules were administered per os with 200 ml still mineral water. We used 5, 10, 20, and 100 mg caffeine-filled capsules. The average body weight was 70.52 kg (SEM: 3.66) and the average caffeine dose was 211.56 mg (SEM: 10.98).” | mechanisms[9] |
| mechanism | identical capsule counts in placebo and caffeine sessions, to stop participants inferring their condition from the number swallowed — “To avoid possible influences caused by subjective bias on the number of capsules taken, volunteers received the same amount of capsules in the control (placebo) sessions as in the caffeine sessions.” | mechanisms[10] |
| mechanism | only non-smokers enrolled, because smoking cuts caffeine half-life by 30 to 50% — “Because the half-life of caffeine in the body is reduced by 30 to 50% in smokers compared to nonsmokers 13 , here we enrolled only nonsmokers.” | mechanisms[11] |
| replicates | Foxe et al. (reference 24), who found with a go/no-go task that caffeine decreased pre-target alpha power — “In the present study we used a simple visual oddball paradigm and we also found that caffeine decreased the pre-target alpha power both in Low and High target probability conditions.” | claim_relations[0] |
| replicates | Gomez and co-workers (reference 36), who reported reduced pre-target gamma power speeding up processing of the forthcoming target — “In line with their results, in the present study, in the pre-target period, we found lower gamma amplitude with faster RT in the High local probability condition and higher gamma amplitude with slower RT in the Low local probability condition.” | claim_relations[1] |
| extends | Trunk et al. (reference 5), the authors' own earlier P300/ERP analysis of the same dataset — “These null effects of UMTS exposure on visual discrimination are in line our previous report on the same behavioral dataset 5 where we found that the UMTS exposure had no observable modulatory effects on RT and P300 ERP measures either alone or in combination with caffeine.” | claim_relations[2] |
| confirms | previous studies finding no effect of UMTS exposure on reaction time — “In line with the results of previous studies 50 , 51 we found no evidence that UMTS exposure alters RT.” | claim_relations[3] |
| supports | the established finding that caffeine decreases reaction time and reduces alpha power — “Tukey HSD post hoc analysis revealed significant amplitude decrease of both alpha bands [alpha1 (p < 0.005), alpha2 (p < 0.005)] in the caffeine treatment compared to control.” | claim_relations[4] |
| contradicts | the authors' own prior hypothesis of a synergistic caffeine-by-EMF effect — “Our hypothesis was that, due to previously reported 1 , 6 similar facilitatory effects on brain excitatory activity, simultaneous caffeine and MP exposure will have a larger effect than caffeine or MP EMF exposure alone.” | claim_relations[5] |
| qualifies | its own implicit short-term memory interpretation, on the basis of the regression results — “However, the results of the multiple regression models showed no interaction effects between the target Probability and Treatments which support a different possible explanation of the present results. Namely, caffeine only facilitates the faster initiation of an already prepared response, independently of the underlying differential cognitive processes (e.g., higher or lower stimulus expectancy).” | claim_relations[6] |
| sense | Low and High probability are stipulated as exactly one versus exactly four preceding standards, on the ground that prior work found no further change beyond four — “We used the terms of Low and High because prior studies have shown no difference between the RTs to targets if the number of the preceding standard stimuli were more than four 41 . Thus Low (Prob1) and High (Prob4) can be considered as most representative categories to describe the possible involvement of different target expectancy levels in task performance.” | senses[0] |
| sense | pharmacologically validated means the study contains an agent known to produce an effect, so a null on the other agent is a null in a demonstrably sensitive assay — “Caffeine exposure also served as pharmacological (positive) control as it has been previously reported that intraoral administration of 3 mg/kg b.w. caffeine reliably improved cognitive functions and mood” | senses[1] |
| sense | synergy is operationalised as a violation of additivity - the sum of single-agent effects against the combined condition, not as a simple interaction term — “To reveal potential interactions of caffeine and UMTS MP exposure in RT and oscillatory measures we focused on the exposure block and we used the additive analysis model with the following formula, as applied elsewhere 42 , 43 , 44 : Hereafter, “[Caffeine − Control] + [UMTS − Control]” and “Combined − Control” are referred to as ‘sum’ and ‘simultaneous’ data, respectively. We hypothesized that violation of the additivity of the analyzed RT or spectral amplitude measures would indicate synergistic interactions.” | senses[2] |
| control | matching the NUMBER of capsules between placebo and caffeine sessions, not merely their contents — held — rules out the participant inferring their condition from how many capsules they were given, and the subjective bias that would follow — “To avoid possible influences caused by subjective bias on the number of capsules taken, volunteers received the same amount of capsules in the control (placebo) sessions as in the caffeine sessions.” | controls[0] |
| control | a two-position switch on the RF amplifier setting genuine or sham exposure, so neither party could tell which was running - the threat is named only by the word blind, which is why this is scored lower — held — rules out expectancy on the part of the participant or the experimenter about which exposure condition was in force — “Double-blind experimental conditions were ensured by a two-position switch (A or B) located on the front panel of the RF amplifier: one position was associated with genuine exposure, and the other with sham exposure.” | controls[1] |
| differs in | its own implicit short-term memory interpretation, on the basis of the regression results — THE AXIS IS WHETHER THE DRUG ACTS ON PREPARATION OR ON PROCESSING. 'the results of the multiple regression models showed no interaction effects between the target Probability and Treatments which support a different possible explanation of the present results. Namely, caffeine only facilitates the faster initiation of an already prepared response, independently of the underlying differential cognitive processes (e.g., higher or lower stimulus expectancy).' The paper withdraws its own cognitive reading on the strength of a NULL INTERACTION - the absence of a statistical interaction is doing the work, and the alternative it licenses is a motor account rather than a memory one. — “However, the results of the multiple regression models showed no interaction effects between the target Probability and Treatments which support a different possible explanation of the present results. Namely, caffeine only facilitates the faster initiation of an already prepared response, independently of the underlying differential cognitive processes (e.g., higher or lower stimulus expectancy).” — “the results of the multiple regression models showed no interaction effects between the target Probability and Treatments which support a different possible explanation of the present results. Namely, caffeine only facilitates the faster initiation of an already prepared response, independently of the underlying differential cognitive processes (e.g., higher or lower stimulus expectancy).” | claim_relations[6] |
| effect direction | caffeine —causes→ reaction time — reduces — “The Tukey HSD post hoc test indicated that the High RT shortened in the caffeine (p < 0.039) and combined (p < 0.023) treatments compared to control” — “shortened” | concept_edges[0] |
| note | caffeine —causes→ reaction time — SCOPE: the effect appears only in High local probability trials; the overall treatment effect did not survive post hoc testing — “The Tukey HSD post hoc test indicated that the High RT shortened in the caffeine (p < 0.039) and combined (p < 0.023) treatments compared to control” | concept_edges[0] |
| weight | caffeine —causes→ reaction time — 0.85 — “The Tukey HSD post hoc test indicated that the High RT shortened in the caffeine (p < 0.039) and combined (p < 0.023) treatments compared to control” | concept_edges[0] |
| effect direction | caffeine —causes→ alpha — reduces — “Tukey HSD post hoc analysis revealed significant amplitude decrease of both alpha bands [alpha1 (p < 0.005), alpha2 (p < 0.005)] in the caffeine treatment compared to control.” — “significant amplitude decrease of both alpha bands” | concept_edges[1] |
| weight | caffeine —causes→ alpha — 0.95 — “Tukey HSD post hoc analysis revealed significant amplitude decrease of both alpha bands [alpha1 (p < 0.005), alpha2 (p < 0.005)] in the caffeine treatment compared to control.” | concept_edges[1] |
| audit | local probability —causes→ reaction time — Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span reports that low- and high-probability categories differed; causes is loose. — “The analysis of the Low (mean: 442.22 ms, SEM: 10.16) and High (mean: 398.54 ms, SEM: 9.71) probability categories revealed that they significantly differed from each other [F(1,22) = 102.584, p < 0.001, eta-squared = 0.823].” | concept_edges[3] |
| outcome (stated by the tagger, no quotation) | control — held | controls[0] |
| outcome (stated by the tagger, no quotation) | control — held | controls[1] |
| for which use | limitation — the null UMTS finding: it cannot be separated from a failed exposure, because THE APPLIED SIGNAL MODULATION WAS INEFFECTIVE OR THE SIGNAL INTENSITY WAS TOO LOW, I.E., UNDER THE THRESHOLD FOR DETECTABLE BIOLOGICAL EFFECTS — “A possible explanation for lack of UMTS exposure effects may be that the applied signal modulation was ineffective or the signal intensity was too low, i.e., under the threshold for detectable biological effects.” | limitations[0] |
| for which use | limitation — the cognitive account of the caffeine benefit: FUTURE STUDIES WITH MORE FOCUSED TASK DESIGN SHOULD ADDRESS THIS QUESTION — “However, to draw the final conclusion we suggest that future studies with more focused task design should address this question.” | limitations[1] |
| for which use | limitation — the overall reaction-time effect: THE TREATMENT EFFECT DID NOT SURVIVE THE POST HOC ANALYSES — “However the Treatment effect did not survive the post hoc analyses.” | limitations[2] |
| for which use | limitation — the paper's own preferred interpretation: THE MULTIPLE REGRESSION MODELS SHOWED NO INTERACTION EFFECTS BETWEEN THE TARGET PROBABILITY AND TREATMENTS WHICH SUPPORT A DIFFERENT POSSIBLE EXPLANATION - caffeine facilitating an already prepared response — “However, the results of the multiple regression models showed no interaction effects between the target Probability and Treatments which support a different possible explanation of the present results. Namely, caffeine only facilitates the faster initiation of an already prepared response, independently of the underlying differential cognitive processes (e.g., higher or lower stimulus expectancy).” | limitations[3] |
| for which use | limitation — the same scope as limitations[1] - the mechanism is not decided by this task design — “However, to draw the final conclusion we suggest that future studies with more focused task design should address this question.” | limitations[4] |
| for which use | limitation — the sample every EEG figure rests on: FOUR PARTICIPANTS WERE EXCLUDED ... THE FINAL SAMPLE IN THE SPECTRAL AMPLITUDE ANALYSIS COMPRISED DATA OF 21 PARTICIPANTS — “Due to excessive artifacts or data loss, four participants were excluded from further analysis. The final sample in the spectral amplitude analysis comprised data of 21 participants (11 female, mean age 20.48 years, SD 1.4).” | limitations[5] |
| for which use | limitation — any count of independent studies: WE ANALYZED THE RECORDED DATA IN CONJUNCTION WITH A PREVIOUS STUDY ... IN A DIFFERENT ASPECT - the same dataset, reported twice — “Here we analyzed the recorded data in conjunction with a previous study 5 in a different aspect. In a previous paper we analyzed the potential effects of caffeine and EMF on stimulus-evoked brain potentials (P300). Here we focus on spectral power of pre-target oscillatory activity because several studies found that caffeine and EMF alters brain oscillations.” | limitations[6] |
| for which use | limitation — the combined-exposure hypothesis stated at the outset: THE PRESENT RESULTS DO NOT SUPPORT THE NOTION THAT UMTS EXPOSURE MAY STRENGTHEN THE OBSERVED FACILITATORY EFFECTS OF CAFFEINE ... IN A COMBINED OR SYNERGISTIC MANNER — “In addition, the present results do not support the notion that UMTS exposure may strengthen the observed facilitatory effects of caffeine on RT in a combined or synergistic manner” | limitations[7] |
| replication | untested | |
| limitation | The null UMTS result may reflect an ineffective modulation or a signal intensity below the threshold for biological effects rather than a true absence of any mobile-phone effect. — “A possible explanation for lack of UMTS exposure effects may be that the applied signal modulation was ineffective or the signal intensity was too low, i.e., under the threshold for detectable biological effects.” | limitations[0] |
| limitation | The task design cannot decide between the two competing explanations of the caffeine benefit (increased sensitivity to the stimulus pattern versus faster initiation of an already prepared response). — “However, to draw the final conclusion we suggest that future studies with more focused task design should address this question.” | limitations[1] |
| limitation | the treatment main effect on overall reaction time did not survive post hoc testing — “However the Treatment effect did not survive the post hoc analyses.” | limitations[2] |
| limitation | the paper's own preferred interpretation is undercut by its regression models, which show no probability-by-treatment interaction — “However, the results of the multiple regression models showed no interaction effects between the target Probability and Treatments which support a different possible explanation of the present results. Namely, caffeine only facilitates the faster initiation of an already prepared response, independently of the underlying differential cognitive processes (e.g., higher or lower stimulus expectancy).” | limitations[3] |
| limitation | the authors call for a more focused task design before concluding — “However, to draw the final conclusion we suggest that future studies with more focused task design should address this question.” | limitations[4] |
| limitation | attrition: two participants dropped from the reaction time analysis and four from the EEG analysis — “Due to excessive artifacts or data loss, four participants were excluded from further analysis. The final sample in the spectral amplitude analysis comprised data of 21 participants (11 female, mean age 20.48 years, SD 1.4).” | limitations[5] |
| limitation | the same dataset was already analysed in a previous paper for evoked potentials — “Here we analyzed the recorded data in conjunction with a previous study 5 in a different aspect. In a previous paper we analyzed the potential effects of caffeine and EMF on stimulus-evoked brain potentials (P300). Here we focus on spectral power of pre-target oscillatory activity because several studies found that caffeine and EMF alters brain oscillations.” | limitations[6] |
| limitation | the hypothesis stated at the outset - that combined exposure would exceed either alone - was not supported — “In addition, the present results do not support the notion that UMTS exposure may strengthen the observed facilitatory effects of caffeine on RT in a combined or synergistic manner” | limitations[7] |
finding · quantity · variable · definition · key-term — grounded in the full text| finding | Reaction times were significantly faster to targets preceded by four standards (High local probability) than to targets preceded by one standard (Low local probability). — “In the present study RTs were highly dependent on the number of standard stimuli preceding the targets, namely RTs significantly decreased for targets with higher local probability (High, preceded by 4 standards) compared to low probability targets (Low, preceded by 1 standard).” | content_tags.findings[0] |
| finding | Caffeine significantly shortened reaction time for High local-probability targets relative to placebo. — “Caffeine treatment significantly decreased the High probability RT relative to the Control (placebo).” | content_tags.findings[1] |
| finding | Caffeine reduced pre-target alpha amplitude in both the Low and the High probability conditions relative to placebo. — “Caffeine significantly decreased both Low and High probability Alpha1 amplitudes relative to the Control.” | content_tags.findings[2] |
| finding | Pre-target gamma amplitude was lower when the target was highly expected and faster responses accompanied that lower gamma. — “In line with their results, in the present study, in the pre-target period, we found lower gamma amplitude with faster RT in the High local probability condition and higher gamma amplitude with slower RT in the Low local probability condition.” | content_tags.findings[3] |
| finding | UMTS mobile-phone-like exposure did not change reaction time. — “In line with the results of previous studies 50 , 51 we found no evidence that UMTS exposure alters RT.” | content_tags.findings[4] |
| finding | UMTS exposure did not change pre-target alpha or gamma power either. — “The present results showing no effect of UMTS MP-like exposure on the alpha and gamma power in the pre-target period correspond well with previous studies using resting EEG, where also no effects of UMTS exposure were reported on brain oscillations” | content_tags.findings[5] |
| finding | Caffeine and UMTS exposure did not interact: the combined condition produced no effect beyond the sum of the single exposures, on either behaviour or oscillations. — “Furthermore, we found no evidence of any interaction between caffeine and UMTS exposure on either RT or brain oscillations (spectral amplitude) using rANOVA or the additive analysis models.” | content_tags.findings[6] |
| finding | The multiple regression models found no Probability-by-Treatment interaction, which supports a different explanation of the results. — “However, the results of the multiple regression models showed no interaction effects between the target Probability and Treatments which support a different possible explanation of the present results.” | content_tags.findings[7] |
| finding | reaction time fell as local target probability rose, fitting a logarithmic model with R-squared 0.88 — “The linear-log statistical model [y = a*log(x) + b] showed significantly decreasing (p < 0.001) logarithmic correlation (R 2 = 0.8832) between the target-target probabilities and the reaction times.” | content_tags.findings[9] |
| finding | High probability trials were about 44 ms faster than Low — “The analysis of the Low (mean: 442.22 ms, SEM: 10.16) and High (mean: 398.54 ms, SEM: 9.71) probability categories revealed that they significantly differed from each other [F(1,22) = 102.584, p < 0.001, eta-squared = 0.823].” | content_tags.findings[10] |
| finding | caffeine shortened High-probability reaction time relative to control — “The Tukey HSD post hoc test indicated that the High RT shortened in the caffeine (p < 0.039) and combined (p < 0.023) treatments compared to control” | content_tags.findings[11] |
| finding | caffeine reduced both alpha bands relative to placebo — “Tukey HSD post hoc analysis revealed significant amplitude decrease of both alpha bands [alpha1 (p < 0.005), alpha2 (p < 0.005)] in the caffeine treatment compared to control.” | content_tags.findings[12] |
| finding | UMTS exposure alone or combined altered neither reaction time nor pre-stimulus oscillations — “However, in the present pharmacologically validated study UMTS exposure either alone or in combination with caffeine did not alter RT or pre-stimulus oscillatory brain activity.” | content_tags.findings[13] |
| finding | the additive model showed no synergy on reaction time — “The summative analysis did not show any synergistic interactions on the RT.” | content_tags.findings[14] |
| finding | the additive model showed no synergy on either alpha band — “The rANOVA [(sum vs. simultaneous) X Probability X Electrode] of the summative model indicated no synergistic interactions of caffeine and UMTS exposure on either alpha1 or alpha2 frequency bands.” | content_tags.findings[15] |
| finding | the overall treatment effect on reaction time did not survive post hoc testing — “However the Treatment effect did not survive the post hoc analyses.” | content_tags.findings[16] |
| quantity | 25 — participants who completed the full factorial caffeine x UMTS design — “In a full factorial experimental design, 25 participants performed a simple visual target detection task while reaction time (RT) and electroencephalogram (EEG) was recorded.” | content_tags.quantities[0] |
| quantity | 23 — participants retained in the reaction-time analysis after exclusions — “The final sample in the RT analysis comprised data from 23 participants (13 female, mean age 20.35 years, SD 1.4).” | content_tags.quantities[1] |
| quantity | 21 — participants retained in the spectral amplitude analysis after artefact-related exclusions — “The final sample in the spectral amplitude analysis comprised data of 21 participants (11 female, mean age 20.48 years, SD 1.4).” | content_tags.quantities[2] |
| quantity | Three mg/kg — caffeine dose administered per kilogram of body weight in the active conditions — “Three mg/kg caffeine packed in identical hard gelatin capsules were administered to the participants.” | content_tags.quantities[3] |
| quantity | 640 and 160 — number of standard and deviant (target) trials presented per session — “The trial numbers for standard and deviant stimuli were 640 and 160, respectively.” | content_tags.quantities[4] |
| quantity | 30 to 50% — reduction in caffeine half-life in smokers, the reason only non-smokers were enrolled — “Because the half-life of caffeine in the body is reduced by 30 to 50% in smokers compared to nonsmokers 13 , here we enrolled only nonsmokers.” | content_tags.quantities[5] |
| quantity | 25 participants, 9 female, mean age 21.07 — the enrolled sample — “Twenty-five healthy, right-handed, non-smoker university students [9 female, age range 18 to 38 years, mean 21.07, standard deviation (SD) 3.58] participated in the study, who regularly consume 1” | content_tags.quantities[6] |
| quantity | 23 in the reaction time analysis, 21 in the spectral analysis — after exclusions for accuracy, data loss and artefacts — “Due to excessive artifacts or data loss, four participants were excluded from further analysis. The final sample in the spectral amplitude analysis comprised data of 21 participants (11 female, mean age 20.48 years, SD 1.4).” | content_tags.quantities[7] |
| value source note | quantity — THE VALUE DRAWS ON THE PAPER BEYOND THIS QUOTE: part of it is printed elsewhere in the paper, not in this quote ('sample in the RT analysis comprised data from 23 participants (13 female, mean age 20.35 years, SD'). | content_tags.quantities[7] |
| quantity | 3 mg/kg, averaging 211.56 mg — the individualised caffeine dose — “Three mg/kg caffeine packed in identical hard gelatin capsules were administered to the participants. The capsules were administered per os with 200 ml still mineral water. We used 5, 10, 20, and 100 mg caffeine-filled capsules. The average body weight was 70.52 kg (SEM: 3.66) and the average caffeine dose was 211.56 mg (SEM: 10.98).” | content_tags.quantities[8] |
| quantity | Low 442.22 ms versus High 398.54 ms — the probability effect on reaction time — “The analysis of the Low (mean: 442.22 ms, SEM: 10.16) and High (mean: 398.54 ms, SEM: 9.71) probability categories revealed that they significantly differed from each other [F(1,22) = 102.584, p < 0.001, eta-squared = 0.823].” | content_tags.quantities[9] |
| quantity | R-squared 0.8832 — the log fit of reaction time to target-target distance — “The linear-log statistical model [y = a*log(x) + b] showed significantly decreasing (p < 0.001) logarithmic correlation (R 2 = 0.8832) between the target-target probabilities and the reaction times.” | content_tags.quantities[10] |
| quantity | alphaI 8-10, alphaII 11-13, gamma1 33-46, gamma2 54-70 Hz — the pre-defined analysis bands — “Spectral amplitudes were calculated for pre-defined frequency bands for statistical analysis (alphaI: 8–10 Hz, alphaII: 11–13 Hz, gamma1: 33–46 Hz, gamma2: 54–70 Hz).” | content_tags.quantities[11] |
| quantity | F(1,20) = 0.001, p = 0.98 and F(3,60) = 0.34, p = 0.79 — trial-count comparison across probability and treatment - the statistics reported in a sentence whose prose asserts significant differences — “There were significant differences between the analyzed trial numbers across the Probability categories (Low, High) [F(1,20) = 0.001; p = 0.98; partial eta-squared < 0.01] or Treatment (Control, UMTS, Caffeine, Combined) [F(3,60) = 0.34; p = 0.79; partial eta-squared < 0.02].” | content_tags.quantities[12] |
| variable | caffeine treatment (independent) — “Three mg/kg caffeine packed in identical hard gelatin capsules were administered to the participants.” | content_tags.variables[0] |
| variable | RF exposure (independent) — “Double-blind experimental conditions were ensured by a two-position switch (A or B) located on the front panel of the RF amplifier: one position was associated with genuine exposure, and the other with sham exposure.” | content_tags.variables[1] |
| variable | reaction time (dependent) — “In a full factorial experimental design, 25 participants performed a simple visual target detection task while reaction time (RT) and electroencephalogram (EEG) was recorded.” | content_tags.variables[3] |
| variable | alpha-band power (dependent) — “We analyzed single trial RT and alpha-band power (amplitude) in the pre-target interval.” | content_tags.variables[4] |
| variable | number of capsules taken (control) — “To avoid possible influences caused by subjective bias on the number of capsules taken, volunteers received the same amount of capsules in the control (placebo) sessions as in the caffeine sessions.” | content_tags.variables[6] |
| variable | target probability (moderator) — “The interaction test between the Probability and the Treatments allows us to investigate whether or not the caffeine effect was modulated by the target probability.” | content_tags.variables[9] |
| variable | target-to-target distance — “Data from Prob1 trials where only one standard stimulus preceded the target were assigned to the low probability (Low) category while data from Prob4 trials where four standard stimuli preceded the target were assigned to the high probability category (High) and were selected for further analysis.” | content_tags.variables[10] |
| variable | pre-target spectral amplitude — “Spectral amplitudes were calculated for pre-defined frequency bands for statistical analysis (alphaI: 8–10 Hz, alphaII: 11–13 Hz, gamma1: 33–46 Hz, gamma2: 54–70 Hz).” | content_tags.variables[11] |
| variable | caffeine concentration — “Saliva samples were taken at the beginning and the end of each recording session and caffeine concentrations were determined by high-performance liquid chromatography (HPLC).” | content_tags.variables[12] |
| definition | probability category — “In the Low probability category and in the High probability category 1 and 4 standard stimuli preceded the target, respectively.” | content_tags.definitions[0] |
| definition | synergistic interaction — “We hypothesized that violation of the additivity of the analyzed RT or spectral amplitude measures would indicate synergistic interactions.” | content_tags.definitions[1] |
| definition | local probability — “The procedure resulted in 7 different stimulus categories according to target local probability (Prob) from category 1 to category 7.” | content_tags.definitions[2] |
| definition | additive analysis model — “To reveal potential interactions of caffeine and UMTS MP exposure in RT and oscillatory measures we focused on the exposure block and we used the additive analysis model with the following formula, as applied elsewhere 42 , 43 , 44 : Hereafter, “[Caffeine − Control] + [UMTS − Control]” and “Combined − Control” are referred to as ‘sum’ and ‘simultaneous’ data, respectively. We hypothesized that violation of the additivity of the analyzed RT or spectral amplitude measures would indicate synergistic interactions.” | content_tags.definitions[4] |
| key-terms | caffeine, UMTS, mobile phone, reaction time, alpha, gamma, oddball, local probability, arousal, HPLC, saliva, placebo, additive, electroencephalogram, pre-target, synergistic, adenosine, target detection, spectral amplitude, full factorial |
| ANOVA F | as written: repeated measures analysis of variance (rANOVA) — “As data followed a normal distribution (determined by Shapiro-Wilk tests), repeated measures analysis of variance (rANOVA) was used to compare the means between groups.” | content_tags.statistics[0] |
| Shapiro-Wilk normality test | as written: Shapiro-Wilk tests — “As data followed a normal distribution (determined by Shapiro-Wilk tests), repeated measures analysis of variance (rANOVA) was used to compare the means between groups.” | content_tags.statistics[1] |
| Tukey HSD post hoc test | as written: post hoc Tukey’s honestly significant difference (HSD) — “Where main effects or interactions were found, statistical results were further specified by post hoc Tukey’s honestly significant difference (HSD).” | content_tags.statistics[2] |
| effect size partial eta-squared | as written: partial eta- squared estimates of effect size — “Each P-value with partial eta- squared estimates of effect size are given in the Results Section.” | content_tags.statistics[3] |
| statistical-significance α | as written: significance level of 0.05 (alpha) — “The null hypothesis was rejected at a significance level of 0.05 (alpha). Each P-value with partial eta- squared estimates of effect size are given in the Results Section.” | content_tags.statistics[4] |
| t-test | as written: Student’s t-test — “We also analyzed possible probability effects with Student’s t-test in each Treatment condition separately.” | content_tags.statistics[5] |
| correlation tau | as written: Kendall’s correlation — “Kendall’s correlation also showed that RT marginally decreased” | content_tags.statistics[6] |
| regression β | as written: Multiple regression statistical method — “Multiple regression statistical method was used to address the question how the different treatments contributed to behavioral effects observed in RT measures controlling for probability categories, and pre-target EEG amplitudes.” | content_tags.statistics[7] |
| mean SD | as written: mean 21.07, standard deviation (SD) 3.58 — “Twenty-five healthy, right-handed, non-smoker university students [9 female, age range 18 to 38 years, mean 21.07, standard deviation (SD) 3.58] participated in the study, who regularly consume 1–2 cups of tea/coffee by self-report.” | content_tags.statistics[8] |
| repeated measures ANOVA with Tukey HSD | — “The Tukey HSD post hoc test indicated that the High RT shortened in the caffeine (p < 0.039) and combined (p < 0.023) treatments compared to control” | content_tags.statistics[9] |
| additive model of sum versus simultaneous | — “To reveal potential interactions of caffeine and UMTS MP exposure in RT and oscillatory measures we focused on the exposure block and we used the additive analysis model with the following formula, as applied elsewhere 42 , 43 , 44 : Hereafter, “[Caffeine − Control] + [UMTS − Control]” and “Combined − Control” are referred to as ‘sum’ and ‘simultaneous’ data, respectively. We hypothesized that violation of the additivity of the analyzed RT or spectral amplitude measures would indicate synergistic interactions.” | content_tags.statistics[10] |
| partial eta-squared | — “There were significant differences between the analyzed trial numbers across the Probability categories (Low, High) [F(1,20) = 0.001; p = 0.98; partial eta-squared < 0.01] or Treatment (Control, UMTS, Caffeine, Combined) [F(3,60) = 0.34; p = 0.79; partial eta-squared < 0.02].” | content_tags.statistics[11] |
| linear-log model | — “The linear-log statistical model [y = a*log(x) + b] showed significantly decreasing (p < 0.001) logarithmic correlation (R 2 = 0.8832) between the target-target probabilities and the reaction times.” | content_tags.statistics[12] |
| method | full factorial within-subject design with placebo, caffeine, UMTS exposure and both, in an active visual oddball task with EEG and saliva caffeine assay — a second reading DIFFERING from the recorded value; recorded: electrophysiology — “In a full factorial experimental design, 25 participants performed a simple visual target detection task while reaction time (RT) and electroencephalogram (EEG) was recorded.” | alternatives.method |
| phenomenon | local target probability processing, indexed by reaction time and pre-target oscillations — a second reading DIFFERING from the recorded value; recorded: caffeine and mobile-phone RF exposure effects on visual target detection — “We found that RT was shorter in High vs. Low local probability trials, and caffeine further shortened RT in High probability trials relative to the baseline condition suggesting that caffeine improves the efficiency of implicit short-term memory.” | alternatives.phenomenon |
Full text — tagged
The reference list is not part of the text we hold for this paper. Our copy was extracted from the publisher's page and the bibliography did not survive that extraction, so the article below ends before its references. Nothing has been removed by us.
Effects of concurrent caffeineterm and mobile phoneterm·key exposure on local target probabilityterm·var processing in the human brain Attila TrunkpersName Attila TrunkpersName 1 Department of Experimental Neurobiology, University of Pécs, Hungary 1 , Gábor StefanicspersName Gábor StefanicspersName 2 Translational Neuromodeling Unit (TNU), Institute for Biomedical Engineering, University of Zurich & ETH Zurich, Switzerland 3 Laboratory for Social and Neural Systems Research, Department of Economics, University of Zürich, Switzerland 2, 3 , Norbert ZentaipersName Norbert ZentaipersName 1 Department of Experimental Neurobiology, University of Pécs, Hungary 1 , Ivett BacskaypersName Ivett BacskaypersName 4 Department of Analytical and Environmental Chemistry, University of Pécs, Hungary 5 Szentágothai Research Centre, University of Pécs, Hungary 4, 5 , Attila FelingerpersName Attila FelingerpersName 4 Department of Analytical and Environmental Chemistry, University of Pécs, Hungary 4 , György ThuróczypersName György ThuróczypersName 6 National Institute for Radiobiology and Radiohygiene (NIRR), Budapest, Hungary 6 , István HernádipersName István HernádipersName 1 Department of Experimental Neurobiology, University of Pécs, Hungary 5 Szentágothai Research Centre, University of Pécs, Hungary 1, 5, a 1 Department of Experimental Neurobiology, University of Pécs, Hungary 2 Translational Neuromodeling Unit (TNU), Institute for Biomedical Engineering, University of Zurich & ETH Zurich, Switzerland 3 Laboratory for Social and Neural Systems Research, Department of Economics, University of Zürich, Switzerland 4 Department of Analytical and Environmental Chemistry, University of Pécs, Hungary 5 Szentágothai Research Centre, University of Pécs, Hungary 6 National Institute for Radiobiology and Radiohygiene (NIRR), Budapest, Hungary a Email: hernadi@ttk.pte.hu Received 2015 May 19; Accepted 2015 Aug 28; Collection date 2015.
Copyright © 2015, Macmillan Publishers Limited This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ PMCID: PMC4585767 PMID: 26395526 Abstract Millions of people use mobile phonesterm·key (MP) while drinking coffee or other caffeineterm containing beverages. Little is known about the potential combined effects of MP irradiation and caffeineterm on cognitive functions. Here we investigated whether caffeineterm intake and concurrent exposure to Universal Mobile Telecommunications System (UMTSterm) MP-like irradiation may interactively influence neuro-cognitive function in an active visual oddballterm paradigm. In a full factorialterm·key experimental design, 25 participants performed a simple visual target detectionterm·key task while reaction timeterm (RT) and electroencephalogramterm·key (EEG) was recorded. Target trials were divided into Low and High probability sets based on target-to-target distanceterm·var. We analyzed single trial RT and alpha-band powerterm·var (amplitude) in the pre-targetterm·key interval. We found that RT was shorter in High vs. Low local probabilityterm trials, and caffeineterm further shortened RT in High probability trials relative to the baseline condition suggesting that caffeineterm improves the efficiency of implicit short-term memoryterm. Caffeineterm also decreased pre-targetterm·key alphaterm amplitude resulting in higher arousalterm level. Furthermore, pre-targetterm·key gammaterm power positively correlated with RT, which may have facilitated target detectionterm·key. However, in the present pharmacologically validated study UMTSterm exposure either alone or in combination with caffeineterm did not alter RT or pre-stimulus oscillatory brain activity.
Millions of people routinely use handheld mobile phonesterm·key (MP). Most of the energy of electromagnetic fields (EMF) emitted by MPs is absorbed in the head of the user and may affect cognitive functions 1 . People often use EMFs emitted by MPs and consume stimulants (e.g., caffeineterm) at the same time without awareness of possible combined effects 2 . Evidences indicate that the combination of caffeineterm and other EMFs, such as light, may alter arousalterm levels and cognitive functions 3 , 4 . However, to date, most available research on human cognition have only investigated the effects of different types of MP exposures or caffeineterm alone without considering their possible additiveterm effects 1 , 2 , 5 .
It is well known that caffeineterm exerts facilitatory effects on human cognition 6 , 7 , 8 , 9 , 10 , 11 , 12 , which are thought to be indirectly brought about by altering calcium channel activation 13 via blocking natural inhibitory effects mediated by adenosineterm·key A1/A2 receptors 14 . Weak EMFs have also been reported to alter intracellular signaling by increasing calcium ion permeability of the cell membrane 15 , 16 or altering the expression of calcium binding proteins 17 , 18 , 19 . While calcium plays an important role in cognitive functions 20 , 21 , 22 , any combined effects of caffeineterm and MP exposure on calcium related mechanisms may affect cognitive performance indexed by reaction timeterm and brain oscillatory activity.
In the present study we focus on the effects of caffeineterm and MP exposure on cognitive information processing indexed by electroencephalographic (EEG) measures of brain function in correlation with behavioral measures of reaction timeterm (RT). Here we focus on analyses of pre-targetterm·key oscillatory activity in the alphaterm and gammaterm frequency bands as they are considered to be neuronal signatures of stimulus processing and the functional basis of perception and cognition 23 .
First, we tested the possible combined effects of caffeineterm and MP exposure on the pre-targetterm·key alphaterm band. Numerous studies investigated the effects of caffeineterm on brain activity in the alphaterm band. Most of them reported that alphaterm activity is affected by caffeineterm, namely caffeineterm decreases the power of resting state alphaterm band indicating increased actual arousalterm state 6 , 11 , 24 , 25 . Several other studies suggested that weak EMFs emitted by MPs may also alter brain oscillatory activities especially in the alphaterm band 1 , 26 , 27 . Alphaterm band itself plays an important role in different mechanisms such as active inhibitory mechanisms 28 or task-dependent cortical processing 29 as well. This frequency band, particularly in the pre-targetterm·key period, is one of the possible determinants of top-down processing which enhances the speed of sensory input detection 30 , 31 .
Second, we measured the possible combined effects on the gammaterm band activity. Oscillatory activity in the gammaterm frequency band is known to facilitate stimulus processing as well 32 . Several studies suggested that gammaterm oscillations play key roles in attention and stimulus expectation. While attention to a stimulus increases the amplitude of gammaterm activity, the expectation of a stimulus decreases it 23 , 33 , 34 . Several studies showed the role of pre-targetterm·key gammaterm activity in determining the speed of RT. For example, positive correlation was found between pre-targetterm·key gammaterm power and RT 35 , showing that lower gammaterm power was associated with faster RT. Thus, the changes of gammaterm activity in the pre-targetterm·key (expectation) period may facilitate the processing of the forthcoming target event 36 .
Here we analyzed the recorded data in conjunction with a previous study 5 in a different aspect. In a previous paper we analyzed the potential effects of caffeineterm and EMF on stimulus-evoked brain potentials (P300). Here we focus on spectral power of pre-targetterm·key oscillatory activity because several studies found that caffeineterm and EMF alters brain oscillations. In the current study, we aimed at investigating the potential effects of caffeineterm and UMTSterm MP exposure on the different local probabilities of the target stimuli indexed by RT and pre-targetterm·key brain oscillations. Specifically, we investigated how RT and pre-targetterm·key alphaterm and gammaterm spectral amplitudeterm·key in different local target probabilityterm·var categories may be affected by caffeineterm, MP exposure or the combination of these two factors.
Our hypothesis was that, due to previously reported 1 , 6 similar facilitatory effects on brain excitatory activity, simultaneous caffeineterm and MP exposure will have a larger effect than caffeineterm or MP EMF exposure alone.
Materials and Methods Participants Twenty-five healthy, right-handed, non-smoker university students [9 female, age range 18 to 38 years, mean 21.07, standard deviation (SD) 3.58] participated in the study, who regularly consume 1–2 cups of tea/coffee by self-report. Because the half-life of caffeineterm in the body is reduced by 30 to 50% in smokers compared to nonsmokers 13 , here we enrolled only nonsmokers. Participants were asked to abstain from any kind of caffeine-containing substances and alcohol at least 10 and 24 hour prior to each session, respectively. All participants gave their written informed consent after the nature of the experiment had been fully explained. The study was conducted according to the ethical principles stated in the Declaration of Helsinki and applicable national guidelines. The protocol of the study was approved by the Ethical Committee of the University of Pécs. Written informed consent was obtained from all volunteers. EEG recordings were carried out at the Psychophysiology Laboratory of the Integrative and Translational Neuroscience Research Group at the University of Pécs, Hungary.
Caffeine concentrationterm·var measurement from salivaterm·key samples Salivaterm·key samples were taken at the beginning and the end of each recording session and caffeine concentrationsterm·var were determined by high-performance liquid chromatography (HPLCterm). Raw salivaterm·key samples were centrifuged for 20 min at 4000 rpm and at 4 °C. About 1.5 to 2 ml supernatants were centrifuged again at 13000 rpm and at 24 °C. About 0.5 to 1 ml of the supernatant was stored at −80 °C for later HPLCterm analysis. (For the details about the HPLCterm analysis, see supplementary data in our previous study by TrunkpersName et al. 5 ).
Caffeine treatmentterm·var Three mg/kg caffeineterm packed in identical hard gelatin capsules were administered to the participants. The capsules were administered per os with 200 ml still mineral water. We used 5, 10, 20, and 100 mg caffeine-filled capsules. The average body weight was 70.52 kg (SEM: 3.66) and the average caffeineterm dose was 211.56 mg (SEM: 10.98). For placeboterm treatment, glucose filled gelatin capsules were used. Placeboterm capsules contained less than 0.5 g glucose per capsule without any additional substance. Similar capsules were used for each treatment. To avoid possible influences caused by subjective bias on the number of capsules takenterm·var, volunteers received the same amount of capsules in the control (placeboterm) sessions as in the caffeineterm sessions.
EEG recording EEG was recorded with a 32-channel BrainAmp amplifier (Brain Products GmbH, Munich, Germany) using silver-silver-chloride (Ag/AgCl) electrodes placed according to the International 10–20 system in an elastic cap (Easycap, Munich, Germany). The nose served as reference and the forehead as ground. An additional electrooculography (EOG) electrode was placed above the right external canthus. The impedance was measured at the beginning of each session and was adjusted to less than 5 kOhm at all electrodes. On-line band-pass filters were used between 0.016 Hz and 450 Hz with an additional notch filter to attenuate power line at 50 Hz. Raw data were digitized at 16 bit at a sampling rate of 1 kHz. Participants were asked to keep their head and eye-movements at minimum during the whole recording session.
UMTSterm exposure device The UMTSterm MP exposure system was previously developed and successfully used in previous studies 2 , 5 , 37 , 38 . The UMTSterm radiofrequency (RF) exposure was administered by means of a standard Nokia 6650 (Nokia, Espoo, Finland) MP via Phoenix Service Software (v. 2005/44_4_120; Nokia, Espoo, Finland) for 15 minutes ( Fig. 1 ). The MP was connected to an external patch antenna, which was mounted on a plastic headset. Double-blind experimental conditions were ensured by a two-position switch (A or B) located on the front panel of the RF amplifier: one position was associated with genuine exposure, and the other with sham exposure. The investigator was not aware of the actual exposure condition. The peak SAR averaged on 1 g tissue was 1.75 W/kg 38 at 2 cm depth from the shell surface of the phantom, and the averaged SAR over 10 g was set below 2 W/kg in any position within the phantom. These values were below the 2 W/kg limit for RF exposureterm·var of the general public as requested by the 1999/519/EC Recommendation. (For the details on the exposure device and conditions, see supplementary data in our previous study by TrunkpersName et al. 5 ).
Figure 1. Schematic drawing of the exposure system.
During the whole EEG recording session the patch antenna was unilaterally placed at a distance of 4 to 5 mm from the right ear above the tragus, mimicking the most frequent normal position of MP in use as reported by the participants. The phone was connected to a 2W RF amplifier and controlled by the Phoenix Service Software (Nokia).
Stimuli and procedure In a double blind, crossover experimental design, the participants took part in four experimental sessions, corresponding to the four possible exposure conditions (Control—placeboterm caffeineterm & sham UMTSterm, UMTSterm—placeboterm caffeineterm & genuine UMTSterm, Caffeineterm—genuine caffeineterm & sham UMTSterm and Combined—genuine caffeineterm & genuine UMTSterm). In the visual oddballterm task, a square as frequent standard (p = 0.8) or a circle as rare deviant (p = 0.2) were presented in a pseudorandom order ( Fig. 2 ). The trial numbers for standard and deviant stimuli were 640 and 160, respectively. Each recording session consisted of three consecutive recording blocks or trials [2.5 min pre exposure block (standard trials: 80; deviant trials: 20), 15 min genuine or sham MP exposure block (standard trials: 480; deviant trials: 120), 2.5 min post exposure block (standard trials: 80; deviant trials: 20)] with no breaks between blocks. During the whole session the patch antenna was unilaterally placed at a distance of 4 to 5 mm from the right ear above the tragus, mimicking the natural position of MP during a call. The stimulus-onset asynchrony (SOA) varied between 1000 and 2000 ms.
Figure 2. Schematic illustration of the experimental design.
In each session dark grey squares were presented as frequent standard (p = 0.8) and a circles as rare deviant (p = 0.2) stimuli on a light grey background. The participants’ task was to press a button on each occurrence of the rare stimulus. Reaction timeterm and pre-targetterm·key EEG activity to the target stimuli were sorted by target-target. For the probability analysis we defined Low and High probability categories. In the Low probability categoryterm·def and in the High probability categoryterm·def 1 and 4 standard stimuli preceded the target, respectively. Stimulus-onset asynchrony was randomized between 1000–2000 ms.
Data analysis Behavioral and EEG data were analyzed off-line on a personal computer using built-in, self-developed scripts and freeware EEGLAB toolbox 39 in the Matlab (MathWorks, Natick, MA) programming environment. To test for the possible acute interaction effects of caffeineterm and MP exposure on reaction timeterm and EEG we analyzed data from the exposure block.
Reaction timeterm and EEG amplitude in the 600 ms interval preceding target onset were binned based on target-target distances. The procedure resulted in 7 different stimulus categories according to target local probabilityterm (Prob) from category 1 to category 7. The local probabilities of these categories in the stimulus sequence were calculated in all conditions (Control, UMTSterm, Caffeineterm and UMTSterm) with the following formula: where T is the number of trials in the analyzed block (T = 120), Ψ counts the number of the targets in the actual (k) Prob category and i increments in each cycle. For example, if 22 Prob1 trials are located in the sequence [Ψ(Prob 1 ) = 22] then the local probabilityterm of the Prob1 is 22/120 = 0.18. If Ψ(Prob 2 ) = 26, then the local probabilityterm of the Prob2 is 26/(120 − 22) = 0.22. The local probabilityterm of the Prob3 is Ψ(Prob 3 )/[120 − (22 + 26)], etc. Figure 3a shows the probabilities of each Prob category.
Figure 3.
( A ) Results for local probabilities in each probability categoryterm·def. The probability of the target as a forthcoming stimulus increases after each standard stimulus is presented before the target. Here, 90% of the stimuli were presented in probability categories 1 to 7. Ten percent of the targets, which were preceded by more than 7 standards (8 to 14), were not analyzed here. ( B ) Results for reaction timeterm (RT) to target stimuli in each probability categoryterm·def. The y = a*log(x) + b linear-log statistical model revealed significant (p < 0.001) logarithmic correlation (R 2 = 0.8832) between the target-target distances (target probabilities) and the reaction timesterm. Kendall’s test showed marginal correlation (tau = −0.619, p = 0.069) on the RTs across the probability categories (1 to 7), with shorter RT in higher probability categories. Furthermore, we found significant difference between High and Low probability categories. Note: for abbreviations see Fig. 2 .
To study treatment effects on sequential stimulus processing, we used a modified version of data separation method by Holm et al. 40 . Data from Prob1 trials where only one standard stimulus preceded the target were assigned to the low probability (Low) category while data from Prob4 trials where four standard stimuli preceded the target were assigned to the high probability categoryterm·def (High) and were selected for further analysis. We used the terms of Low and High because prior studies have shown no difference between the RTs to targets if the number of the preceding standard stimuli were more than four 41 . Thus Low (Prob1) and High (Prob4) can be considered as most representative categories to describe the possible involvement of different target expectancy levels in task performance.
To reveal potential interactions of caffeineterm and UMTSterm MP exposure in RT and oscillatory measures we focused on the exposure block and we used the additive analysis modelterm·def with the following formula, as applied elsewhere 42 , 43 , 44 : Hereafter, “[Caffeineterm − Control] + [UMTSterm − Control]” and “Combined − Control” are referred to as ‘sum’ and ‘simultaneous’ data, respectively. We hypothesized that violation of the additivity of the analyzed RT or spectral amplitudeterm·key measures would indicate synergistic interactionsterm·def. This hypothesis was tested on both RT and pre-target spectral amplitudeterm·var measures as described later.
As data followed a normal distribution (determined by Shapiro-Wilk teststerm·stat), repeated measures analysis of variance (rANOVA)term·stat was used to compare the means between groups. Where main effects or interactions were found, statistical results were further specified by post hoc Tukey’s honestly significant difference (HSD)term·stat. The null hypothesis was rejected at a significance level of 0.05 (alphaterm). Each P-value with partial eta- squared estimates of effect sizeterm·stat are given in the Results Section.
Reaction timeterm We analyzed RT to correctly responded target which occurred between 50 ms and 1000 ms after the stimulus onset 45 . All responses outside this interval, and when the participants made no button press to targets were ignored. Due to the 80% of the acceptable target accuracy level and data loss two participants were excluded from further analysis. The final sample in the RT analysis comprised data from 23 participants (13 female, mean age 20.35 years, SD 1.4). First, possible target probabilityterm·var effects were analyzed on all probability categories (from Prob1 to Prob7) with a two-way rANOVA (Treatment [Control vs. UMTSterm vs. Caffeineterm vs. Combined] X Probability [Prob1 to Prob7]). Hereafter, two probability categories (Prob1 as Low and Prob4 as High) were chosen for further analysis. The possible synergisticterm·key effects on RT were analyzed with two-way rANOVA (Treatment [Control vs. UMTSterm vs. Caffeineterm vs. Combined] X Probability [Low vs. High]). Furthermore, to analyze the possible Treatment effects even more precisely, we divided RT data into Low only and High only categories and applied one-way rANOVA (Treatment [Control vs. UMTSterm vs. Caffeineterm vs. Combined]). We also analyzed possible probability effects with Student’s t-testterm·stat in each Treatment condition separately.
Spectral amplitudesterm·key in the pre-targetterm·key period Continuous EEG data were off-line band-pass filtered between 0.5 Hz and 80 Hz with 50 Hz notch filter. Pre-targetterm·key epochs from 600 ms preceding targets to the stimuli onset (0 ms) were extracted. Since mean SOA was 1500 ms, the analyzed the segments, which contained mostly spontaneous activity and were mostly free from activity evoked by the standard stimulus, which preceded the target ( Fig. 4 ). For off-line artifact rejection purposes all epochs exceeding ±100 uV on any of the electrodes including the EOG electrode or epochs containing incorrect behavioral response were excluded from further analysis. The overall mean analyzed trial number was 17.02 (SEM: 0.14) and average acceptance trial rate (analyzed/presented trials*100) was 92% (SEM: +−9). There were significant differences between the analyzed trial numbers across the Probability categories (Low, High) [F(1,20) = 0.001; p = 0.98; partial eta-squaredterm·stat < 0.01] or Treatment (Control, UMTSterm, Caffeineterm, Combined) [F(3,60) = 0.34; p = 0.79; partial eta-squaredterm·stat < 0.02].
Figure 4. Grand-average event related potentials (ERPs) recorded from Pz electrode site in each treatment (Control, UMTSterm, Caffeineterm and Combined).
ERPs to standard and deviant stimuli (target) are shown in both Low and High probability categories. The Low and High probability standards were the ERPs evoked immediately before the Low and High probability targets, respectively. Note: for abbreviations see Fig. 2 .
Due to excessive artifacts or data loss, four participants were excluded from further analysis. The final sample in the spectral amplitudeterm·key analysis comprised data of 21 participants (11 female, mean age 20.48 years, SD 1.4). Fast Fourier transformation (FFT) was applied on the artifact free, epoched data with 1 Hz resolution to get spectral power values. These values were than transformed with the following formula 24 : Spectral amplitudesterm·key were calculated for pre-defined frequency bands for statistical analysis (alphaI: 8–10 Hz, alphaII: 11–13 Hz, gamma1: 33–46 Hz, gamma2: 54–70 Hz). The possible synergisticterm·key effects on the amplitudes of the Alpha1, Alpha2, Gamma1 and Gamma2 frequency bands were analyzed at posterior (P3, P4, O1, O2, P7, P8, Pz, Cp1, Cp2, Cp5, Cp6) electrode sites with three-way rANOVA (Probability X Treatment X Electrode), respectively 46 . Where Treatment or Probability categoryterm·def main effects were found, analyses were further refined by dividing the data into Low/High only probability or Control/UMTSterm/Caffeineterm/Combined only subgroups, respectively. On the refined data-sets two-way rANOVA (Treatment X Electrode) or rANOVA (ProbabilityCategory X Electrode) were applied.
Multiple regressionterm·stat Multiple regression statistical methodterm·stat was used to address the question how the different treatments contributed to behavioral effects observed in RT measures controlling for probability categories, and pre-targetterm·key EEG amplitudes. The interaction test between the Probability and the Treatments allows us to investigate whether or not the caffeineterm effect was modulated by the target probabilityterm·var. Here we applied the following multiple regressionterm·stat equation: In this equation Y is the predicted variable, X1 … X10 are the predictor variables, β 0 is the intercept, and ε is the error. The terms β 1 … β 10 are the estimated slope parameters, which are used as multipliers for the corresponding X1 … X10 predictor variables and their interactions, respectively.
First, to address the question whether any treatment (UMTSterm, Caffeineterm, or Combined) affects the RT, we applied categorical variables on Treatments where Control treatment served as reference also controlling for probabilities and pre-target spectral amplitudeterm·var as predictor variables. We applied the following multiple regressionterm·stat model: Second, to reveal any potential combined effects of caffeineterm and UMTSterm MP exposures we applied the same regressionterm·stat model, with the categorical variable for Combined treatment serving as the reference this time. The model is as follows: Results Reaction timeterm Overall, the analysis of the 7 probability categories (Prob1 to Prob7) showed significant main effects of Probability [F(6,132) = 33.615, p < 0.001, eta-squared = 0.604] and Treatment [F(3,66) = 3.035, p = 0.035, eta-squared = 0.12]. The linear-log statistical model [y = a*log(x) + b] showed significantly decreasing (p < 0.001) logarithmic correlation (R 2 = 0.8832) between the target-target probabilities and the reaction timesterm. Kendall’s correlationterm·stat also showed that RT marginally decreased (tau = −0.619, p = 0.069) from the Prob1 (mean: 442.22 ms, SEM: 10.16) to Prob7 (mean: 399.17 ms, SEM: 11.15) ( Fig. 3b ).
The analysis of the Low (mean: 442.22 ms, SEM: 10.16) and High (mean: 398.54 ms, SEM: 9.71) probability categories revealed that they significantly differed from each other [F(1,22) = 102.584, p < 0.001, eta-squared = 0.823]. Furthermore, a significant Treatment main effect [F(3,66) = 2.849, p < 0.045, eta-squared = 0.115] was found. However the Treatment effect did not survive the post hoc analyses.
When we separated RTs into Low only and High only probability categories, the analysis revealed significant Treatment main effects only in the High category [F(3,66) = 3.621, p < 0.018, eta-squared: 0.14]. The Tukey HSD post hoc testterm·stat indicated that the High RT shortened in the caffeineterm (p < 0.039) and combined (p < 0.023) treatments compared to control ( Fig. 5 ). The probability effects were analyzed in all Treatments, separately. Student’s t-testterm·stat showed significant differences between Low and High probability categories in each Treatment (all p < 0.001). The summative analysis did not show any synergistic interactionsterm·def on the RT.
Figure 5. Results for reaction timeterm (RT) to target stimuli in each treatment condition (Control, UMTSterm, Caffeineterm, Combined).
Caffeine treatmentterm·var significantly decreased the High probability RT relative to the Control (placeboterm). We found no combined effects of caffeineterm and UMTSterm exposure on RT. Note: *p < 0.05; for abbreviations see Fig. 2 .
Spectral amplitudesterm·key in the pre-targetterm·key period Pre-targetterm·key EEG amplitudes are shown in Fig. 6 .
Figure 6. Log-transformed pre-targetterm·key (−600 to 0 ms) spectral power at the analyzed electrode sites in each treatment and in analyzed each probability categoryterm·def (Low and High).
Note: for abbreviations see Fig. 2 .
Significant Treatment effects were found in alpha1 [F(3,60) = 8.779, p < 0.005, eta-squared = 0.305] and alpha2 [F(3,60) = 5.552, p < 0.005, eta-squared = 0.217] frequency bands. Tukey HSD post hoc analysis revealed significant amplitude decrease of both alphaterm bands [alpha1 (p < 0.005), alpha2 (p < 0.005)] in the caffeine treatmentterm·var compared to control. Furthermore, the alpha1 amplitude significantly decreased in the caffeineterm and in the combined treatment compared to UMTSterm (p < 0.005) or control (p < 0.05), respectively. Treatment main effects were further specified by dividing the data into Low and High probability categories. The effects of caffeineterm on the alpha1 amplitude are shown in Fig. 7 .
Figure 7. Scalp topographic maps of the Low and High probability Alpha1 amplitudes in each treatment.
Colors represent the mean Alpha1 amplitudes in the −600 to 0 ms time period preceding the targets. The electrode sites in the region of interests are marked with bold face. Note: for abbreviations see Fig. 2 .
The rANOVA yielded the following significant main effects: Low Alpha1: F(3,60) = 7.852, p < 0.005, eta-squared = 0.282; High Alpha1: F(3,60) = 5.547, p < 0.005, eta-squared = 0.217; Low Alpha2: F(3,60) = 4.398, p < 0.005, eta-squared = 0.180; High Alpha2: F(3,60) = 4.032, p < 0.05, eta-squared = 0.168. The results of the Tukey HSD post hoc analysis are shown in Fig. 8 . The rANOVA [(sum vs. simultaneous) X Probability X Electrode] of the summative model indicated no synergistic interactionsterm·def of caffeineterm and UMTSterm exposure on either alpha1 or alpha2 frequency bands.
Figure 8.
( A ) Results for Alpha1 amplitudes in each treatment. Caffeineterm significantly decreased both Low and High probability Alpha1 amplitudes relative to the Control. We found no combined effects of caffeineterm and UMTSterm exposure on the Alpha1. ( B ) Results for Alpha2 amplitudes in each treatment. Caffeineterm significantly decreased the High Alpha2 amplitude relative to the Control. We found no combined effects of caffeineterm and UMTSterm exposure on the Alpha2. Note: *p < 0.05; for abbreviations see Fig. 2 .
Significant probability main effects was found in both gamma1 [F(1,20) = 18.731, p < 0.005, eta-squared = 0.484] and gamma2 [F(1,20) = 33.908, p < 0.005, eta-squared = 0.629] bands in the pre-targetterm·key period. We found that gamma1 and gamma2 pre-targetterm·key amplitudes were significantly lower in the High probability categoryterm·def (gamma1: mean = 0.518 uV, SEM: 0.025; gamma2: mean = 0.364 uV, SEM = 0.015) than in the Low probability (gamma1: mean = 0.551 uV, SEM: 0.027; gamma2: mean = 0.396 uV, SEM: 0.015). No specific Treatment effects were found. When we divided the data into separate datasets containing only Low and High probability categories, significant probability main effects were found on all models except in caffeineterm gamma1 and UMTSterm gamma2 conditions.
The rANOVA [(sum vs. simultaneous) X Probability X Electrode] of the summative model indicated no interactive effects of caffeineterm and UMTSterm exposure on either gamma1 or gamma2 frequency bands.
Multiple regressionterm·stat analysis Estimated parameters and their significance levels for the Equation 1 and Equation 2 are shown in the Table 1 .
Table 1. Results for the multiple regressionterm·stat analysis.
Parameters Equation 1 Equation 2 Predictor variables Estimated parameters Significance levels Predictor variables Estimated parameters Significance levels β 0 Intercept 257.69 <0.05 Intercept 229.21 <0.05 β 1 Probability −3.77 0.09 Probability −4.33 <0.05 β 2 Alpha1 −17.8 NS Alpha1 −17.8 NS β 3 Alpha2 −20.74 NS Alpha2 −20.74 NS β 4 Gamma1 565.54 <0.05 Gamma1 565.54 <0.05 β 5 Gamma2 −12.69 NS Gamma2 −12.69 NS β 6 UMTSterm −14.98 NS Control 28.47 0.08 β 7 Caffeineterm −34.5 0.08 UMTSterm 13.49 NS β 8 Combined −28.47 0.08 Caffeineterm −6.03 NS β 9 Probability*UMTSterm −1.92 NS Probability*Control 0.57 NS β 10 Probability*Caffeineterm −2.41 NS Probability*UMTSterm −1.36 NS β 11 Probability*Combined −0.57 NS Probability*Caffeineterm −1.85 NS Estimated parameters and significance levels with predictor variables are listed for Equation 1 and Equation 2 , respectively. In the Equation 1 and Equation 2 , Control and Combined treatments were used as reference variables, respectively.
Discussion In the present study, we investigated possible synergisticterm·key effects of caffeineterm and acute UMTSterm MP exposure on target local probabilityterm indexed by RT and pre-targetterm·key brain activity in a visual target detectionterm·key task, where participants discriminated between frequent standard and rare target stimuli and responded with a button-press to the latter. Caffeineterm exposure also served as pharmacological (positive) control as it has been previously reported that intraoral administration of 3 mg/kg b.w. caffeineterm reliably improved cognitive functions and mood 7 , 9 . To test the possible synergism we adopted an additive analysis modelterm·def 43 , 44 .
Reaction timeterm and brain responses are known to show remarkable individual trial by trial variability depending on the actual arousalterm state or attentional level of the participants 40 . In the present study RTs were highly dependent on the number of standard stimuli preceding the targets, namely RTs significantly decreased for targets with higher local probabilityterm (High, preceded by 4 standards) compared to low probability targets (Low, preceded by 1 standard). Our results are in line with findings in numerous previous studies which investigated the effects of perceived distance to target stimuli by the increasing number of preceding non-targets on RTs and event-related potentials 40 , 41 , 47 . Most of them found that RTs and brain responses to infrequent target stimuli negatively correlated with the number of the frequent non-target stimuli preceding that target. In addition, here we found that caffeineterm further facilitated RT for High local probabilityterm targets, whereas it did not improve RT for targets with Low local probabilityterm. The present results are in line with previous findings showing that caffeineterm decreases RT 6 , 48 , 49 and we also show here that the decreased RT is mainly mediated by responses to highly expected stimuli. One possible explanation of our findings is that caffeineterm speeds up High RTs, which belong to more overt attention. Thus, it would be reasonable to conclude that caffeineterm increases the sensitivity to the stimulus pattern, and improves the efficiency of implicit short-term memoryterm 41 . However, the results of the multiple regressionterm·stat models showed no interaction effects between the target Probabilityterm·var and Treatments which support a different possible explanation of the present results. Namely, caffeineterm only facilitates the faster initiation of an already prepared response, independently of the underlying differential cognitive processes (e.g., higher or lower stimulus expectancy). However, to draw the final conclusion we suggest that future studies with more focused task design should address this question.
In line with the results of previous studies 50 , 51 we found no evidence that UMTSterm exposure alters RT. In addition, the present results do not support the notion that UMTSterm exposure may strengthen the observed facilitatory effects of caffeineterm on RT in a combined or synergisticterm·key manner as results of our additive analysis modelterm·def did not suggest any interactive effects.
Several studies have investigated the role of the alpha-band oscillation during resting conditions and during cognitive task performance and it is widely accepted that the alphaterm frequency band is an important readout of ongoing attentional processes 52 , 53 . For example, using a go/no-go task, Foxe et al. 24 found that caffeineterm decreased alphaterm power in the pre-targetterm·key period. In the present study we used a simple visual oddballterm paradigm and we also found that caffeineterm decreased the pre-targetterm·key alphaterm power both in Low and High target probabilityterm·var conditions. One possible explanation of the decreased alphaterm band activity in the pre-targetterm·key by caffeineterm may be that caffeineterm increased the arousalterm state, thus promoted better visual perception performance compared to the control condition 54 . Alternatively, caffeineterm may have enhanced neural excitability in general 55 . Our results of the multiple regressionterm·stat analysis on alpha1 amplitude and target probabilityterm·var suggest that RT was more influenced by the target probabilityterm·var than by the pre-targetterm·key alphaterm amplitude in all non-caffeine conditions. However, in the caffeineterm condition the RT showed a stronger relationship with the alpha1 amplitude than with target probabilityterm·var. One possible explanation may be that caffeineterm has a general positive effect on general attentional resource allocation independently of the relative target probabilityterm·var 56 .
Several studies investigated the role of gammaterm oscillations in various cognitive functions. For example, pre-targetterm·key gammaterm activity was found sensitive to top-down attention, especially when participants highly anticipated the occurrence of the target stimulus 32 . Elsewhere, oscillation power in the gammaterm band reportedly predicted the speed of RT to the forthcoming stimuli. Reinhart and co-workers tested their hypothesis in an auditory paradigm, where participants had to respond to target tones 35 . The authors found that the decrease of gammaterm power in the pre-targetterm·key period positively correlated with the RT. In the same study, it was also suggested that higher gammaterm power indicated more effective response preparation in the pre-targetterm·key period, as reflected in the more detailed, but much slower evaluation of the forthcoming target stimulus 35 . In another study by Gómez and co-workers the authors found a generalized decrease in the oscillatory activity in the pre-targetterm·key period, and suggested that reduction of the gammaterm power speeds up the processing of the forthcoming target stimulus 36 . In line with their results, in the present study, in the pre-targetterm·key period, we found lower gammaterm amplitude with faster RT in the High local probabilityterm condition and higher gammaterm amplitude with slower RT in the Low local probabilityterm condition. Thus, it is likely that the decreased gammaterm activity prior to the predicted arrival of a target stimulus may facilitate the processing of relevant task-related information 36 .
The present results showing no effect of UMTSterm MP-like exposure on the alphaterm and gammaterm power in the pre-targetterm·key period correspond well with previous studies using resting EEG, where also no effects of UMTSterm exposure were reported on brain oscillations 2 , 37 , 57 , 58 . Thus, our results further support the notion that an acute, 15 min exposure to UMTSterm MP-like EMF signal alone does not affect neural activity concerning decision making in a visual discrimination task. Furthermore, we found no evidence of any interaction between caffeineterm and UMTSterm exposure on either RT or brain oscillations (spectral amplitudeterm·key) using rANOVA or the additive analysis modelsterm·def. In addition, we suggest that 15 min UMTSterm MP-like EMF exposure does not influence (increase or decrease) the observed facilitatory effects of caffeineterm on behavioral or electrophysiological measures of cognitive performance in a visual discrimination task. These null effects of UMTSterm exposure on visual discrimination are in line our previous report on the same behavioral dataset 5 where we found that the UMTSterm exposure had no observable modulatory effects on RT and P300 ERP measures either alone or in combination with caffeineterm.
Conclusion We found that caffeineterm speeds up responses to highly expected targets and facilitates allocation of attentional resources as indexed changes in pre-targetterm·key alphaterm amplitude. Furthermore, pre-targetterm·key gammaterm amplitude negatively correlated with target probabilityterm·var. However, no effects of UMTSterm exposure were observed alone or in combination with caffeineterm, suggesting that UMTSterm exposure did not have any additional facilitatory effect on visual target detectionterm·key. A possible explanation for lack of UMTSterm exposure effects may be that the applied signal modulation was ineffective or the signal intensity was too low, i.e., under the threshold for detectable biological effects. However, at this point, it cannot be generally ruled out, as also suggested by Juutilainen et al. 59 , that other types of frequently used EMF modulation may exceed the threshold for biological effects, either alone or in combination with chemical or other agents (e.g., combined modulation type of EMF). As in the present study and in our previous report 5 we also applied positive pharmacological manipulation to explore possible additiveterm effects of UMTSterm exposure on known facilitatory brain activation in a full factorialterm·key recording design, we generally conclude that an acute 15 min UMTSterm exposure does not alter RT or pre-targetterm·key oscillatory activity.
Additional Information How to cite this article : TrunkpersName, A. et al. Effects of concurrent caffeineterm and mobile phoneterm·key exposure on local target probabilityterm·var processing in the human brain. Sci. Rep. 5 , 14434; doi: 10.1038/srep14434 (2015).
Acknowledgments This work was supported by the Hungarian National Program in Brain Sciences (NPIBS, KTIA NAP 13-1-2013-0001). The project was subsidized by the European Union and co-financed by the European Social Fund. We thank all members of Working Group #3, Part: Human Studies of EU-COST Action BM0704 for valuable discussions on the initial experimental design.
Footnotes Author Contributions I.H., G.S. and A.T. designed research; A.T. and N.Z. performed EEG experiments; I.B. and A.F. performed HPLCterm analyses; G.T. designed exposure system and performed dosimetry; A.T. analyzed data; I.H. G.S. and A.T. wrote the paper. All authors read and approved the final manuscript.