Edited by: Gianfranco Spalletta, Santa Lucia Foundation (IRCCS), Italy
Reviewed by: Daya Shankar Gupta, South University, United States; Francisco Elezier Xavier Magalhães, Federal University of Delta of Parnaíba, Brazil
This article was submitted to Neurodegeneration, a section of the journal Frontiers in Neuroscience
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Cognitive deficits are common in Parkinson’s disease (PD) and range from mild cognitive impairment to dementia, often dramatically reducing quality of life. Physiological models have shown that attention and memory are predicated on the brain’s ability to process time. Perception has been shown to be increased or decreased by activation or deactivation of dopaminergic neurons respectively. Here we investigate differences in time perception between patients with PD and healthy controls. We have measured differences in sub-second- and second-time intervals. Sensitivity and error in perception as well as the response times are calculated. Additionally, we investigated intra-individual response variability and the effect of participant devices on both reaction time and sensitivity. Patients with PD have impaired sensitivity in discriminating between durations of both visual and auditory stimuli compared to healthy controls. Though initially designed as an in-person study, because of the pandemic the experiment was adapted into an online study. This adaptation provided a unique opportunity to enroll a larger number of international participants and use this study to evaluate the feasibility of future virtual studies focused on cognitive impairment. To our knowledge this is the only time perception study, focusing on PD, which measures the differences in perception using both auditory and visual stimuli. The cohort involved is the largest to date, comprising over 800 participants.
Cognitive deficits are common in Parkinson’s disease (PD) and range from mild cognitive impairment to dementia, often dramatically reducing quality of life (
A hallmark feature of PD (
In addition to time perception being used a clinical biomarker of neurodegeneration, measuring changes in awareness can help shed light on pathophysiology (
The aim of this study was to investigate differences in time perception between patients with PD and healthy controls (HCs). Though initially designed as an in-person study, as a consequence of the outbreak of COVID-19 the experiment was adapted into an online-only, virtual study. This adaptation provided a unique opportunity to enroll a larger number of international participants and use this study to evaluate the feasibility of future virtual studies focused on cognitive impairment.
We searched PubMed, Web of Science, and Google Scholar to identify articles published from inception to November 2021, using the keywords “Parkinson’s disease” and “time perception” or “time processing” or “temporal perception,” with English language restrictions. A previous study (
This study aims to measure differences in time perception, using both auditory and visual stimuli, between patients with PD and HCs. We measure differences in sub-second- and second-time intervals. Sensitivity and error in perception as well as the response time are calculated. Additionally, we investigate intra-individual response variability and the effect of participant devices on both reaction time and sensitivity.
To our knowledge this is the only time perception study, focusing on PD, which measures the differences in perception using both auditory and visual stimuli. The cohort involved is the largest to date, comprising over 800 participants.
The initial intention was to conduct this study at in-person visits. However, COVID-19 related travel restrictions led to its adaptation to an online platform. This had unexpected benefits: firstly, the number of possible participants increased substantially, and secondly, an international cohort of participants was able to be recruited. In addition, conducting the study online in participants’ homes using their personal devices, afforded us the opportunity to investigate the feasibility and accuracy of virtual studies using touch screens and home computers. It is also important to highlight the ability of virtual studies to protect participants and remain relatively robust during a long-lasting pandemic whilst still producing coherent data.
The protocol was approved by the University of Oxford Medical Sciences Division Research Ethics Committee (R64730/RE001).
Participants were recruited to PD and HC groups. Participants in the PD group had an established diagnosis of PD with an age at disease onset of over 50 years, had no neurological conditions other than PD, no conditions affecting hearing or vision, and had not undergone deep brain stimulation surgery. HC participants were above 50 years of age without any neurological conditions.
We tested participants’ time perception using both visual and auditory stimuli. The visual stimulus was a clock figure displayed on a screen, of a design inspired by Benjamin Libet’s seminal time perception experiment (
The tasks were completed online using participants’ own devices, which including smartphones, tablet computers, laptops, and desktops. The study was hosted on the Gorilla platform (
Within each group (PD or HC), participants were then randomly assigned to one of two subgroups. One subgroup, referred to as the “subseconds” group, were tested using stimuli with a reference duration of 500 ms for both visual and audio tasks. The other subgroup, referred to as the “seconds” group, were tested using stimuli with a reference duration of 1500 ms for both visual and audio tasks. For each participant, the order of the two tasks (visual then audio or vice versa) was chosen randomly. The auditory task was preceded by a test audio clip and instruction to participants to adjust their devices’ audio levels so that they could hear the audio clip clearly.
Both visual and audio tasks started with two practice trials to familiarize the participant with the task and their response method (keyboard or touch screen response), followed by up to 100 actual trials. Each trial began with a blank screen for 750 ms, followed by a standard stimulus (auditory or visual) of fixed duration (500 ms for the subseconds group and 1,500 ms for the seconds group), then another blank screen for 750 ms, followed by a comparison stimulus of the same modality, with a variable duration that could be longer or shorter than the standard stimulus. Participants were asked to respond by indicating whether the comparison stimulus duration was longer or shorter than the standard stimulus duration.
There were 100 possible comparison stimulus durations ranging from 0.5 to 1.5 times the standard stimulus duration with increments of 1% of the first stimulus duration, and the standard duration was not re-used (i.e., for the subseconds group the comparison stimulus durations ranged from 250 to 750 ms with 5 ms increments excluding 500 ms, and for the seconds group the comparison stimulus durations ranged from 750 to 2,250 ms with 15 ms increments excluding 1,500 ms). Comparison stimulus durations were sampled randomly without replacement from the pool of possible values.
We performed at least 55 trials in each participant; this choice was based upon visual inspection of pilot datasets, in which this was the minimum number required for stability of the estimated parameters. Trials were continued beyond this until the rolling 10-point coefficient of variation of estimates of the logistic regression slope was less than 2% for 5 consecutive trials (see
We calculated the minimum sample size needed for the study using the G*Power software (
For each trial, a response of “shorter” was assigned the value 0 and “longer” the value 1. The responses were then plotted against the duration of the comparison stimulus and fitted with a logistic regression curve (
Sample logistic regression curve for the subseconds interval, demonstrating the spread of the responses, 0 indicates “shorter” and 1 indicates “longer.” The blue dots represent each response. The solid red line represents the fitted logistic regression curse and the dotted red lines indicate the 95% confidence intervals. The yellow cross represents the point of subjective equality where the participant has equal likelihood to choose either “shorter” or “longer” responses. The slope of the curve at this point, and the shift of its 50% point from the standard duration (500 ms), are used to quantify the sensitivity and error of time perception, respectively.
The model equation was:
where
where standard is the duration of the comparison stimuli in ms.
Because the distribution of logistic regression slopes was highly positively skewed in both PD and control groups, a logarithmic transformation was performed to permit parametric analysis.
To ensure participants understood the task, and in order to only include trials where an active discrimination was made, tasks with negative slopes were excluded as well as participants with invalid or missing audio or visual task data. In line with established procedures in reaction time research (
Intraindividual variability (IIV) of reaction time has been speculated as a cognitive predictor of mild cognitive disorders (
Bonferroni correction was used to correct for multiple comparisons using the Mann–Whitney test.
A total of 320 patients with PD were recruited through a multitude of Parkinson’s associations worldwide and 620 HCs above the age of 50 were recruited using Prolific (
Demographics table of Parkinson’s disease (PD) group and healthy control (HC) group performing the seconds or subseconds task.
Subseconds |
Seconds |
Overall |
||||
HC ( |
PD ( |
HC ( |
PD ( |
HC ( |
PD ( |
|
Mean (SD) | 65.6 (5.35) | 66.5 (6.83) | 66.1 (5.54) | 66.5 (7.52) | 65.8 (5.45) | 66.5 (7.21) |
Median (Min, Max) | 65.0 (50.0, 87.0) | 66.0 (50.0, 85.0) | 65.5 (52.0, 86.0) | 67.0 (50.0, 85.0) | 65.0 (50.0, 87.0) | 67.0 (50.0, 85.0) |
Male | 100 (37.2%) | 61 (47.3%) | 127 (42.3%) | 73 (44.8%) | 227 (39.9%) | 134 (45.9%) |
Female | 169 (62.8%) | 68 (52.7%) | 173 (57.7%) | 90 (55.2%) | 342 (60.1%) | 158 (54.1%) |
Left | 35 (13.0%) | 13 (10.1%) | 29 (9.7%) | 19 (11.7%) | 64 (11.2%) | 32 (11.0%) |
Right | 234 (87.0%) | 111 (86.0%) | 269 (89.7%) | 140 (85.9%) | 503 (88.4%) | 251 (86.0%) |
Ambidextrous | 0 (0%) | 5 (3.9%) | 2 (0.7%) | 4 (2.5%) | 2 (0.4%) | 9 (3.1%) |
Asian | 2 (0.7%) | 1 (0.8%) | 4 (1.3%) | 1 (0.6%) | 6 (1.1%) | 2 (0.7%) |
Hispanic or Latino | 2 (0.7%) | 0 (0%) | 1 (0.3%) | 2 (1.2%) | 3 (0.5%) | 2 (0.7%) |
Other (please specify) | 5 (1.9%) | 1 (0.8%) | 5 (1.7%) | 1 (0.6%) | 10 (1.8%) | 2 (0.7%) |
White | 250 (92.9%) | 102 (79.1%) | 284 (94.7%) | 157 (96.3%) | 534 (93.8%) | 259 (88.7%) |
Black or African American | 0 (0%) | 0 (0%) | 3 (1.0%) | 0 (0%) | 3 (0.5%) | 0 (0%) |
Jewish | 0 (0%) | 0 (0%) | 1 (0.3%) | 0 (0%) | 1 (0.2%) | 0 (0%) |
Mean (SD) | NA (NA) | 6.35 (4.88) | NA (NA) | 6.57 (4.65) | NA (NA) | 6.47 (4.74) |
Median (Min, Max) | NA (NA, NA) | 5.00 (0.0137, 24.0) | NA (NA, NA) | 5.00 (0.00800, 28.0) | NA (NA, NA) | 5.00 (0.00800, 28.0) |
0 (0%) | 4 (3.1%) | 0 (0%) | 6 (3.7%) | 0 (0%) | 10 (3.4%) | |
Levodopa | 0 (0%) | 110 (85.3%) | 0 (0%) | 142 (87.1%) | 0 (0%) | 252 (86.3%) |
Others | 0 (0%) | 14 (10.9%) | 0 (0%) | 15 (9.2%) | 0 (0%) | 29 (9.9%) |
Mean (SD) | NA (NA) | 3.21 (3.49) | NA (NA) | 3.33 (3.37) | NA (NA) | 3.28 (3.41) |
Median (Min, Max) | NA (NA, NA) | 2.00 (0, 21.0) | NA (NA, NA) | 3.00 (0, 24.0) | NA (NA, NA) | 2.50 (0, 24.0) |
Computer | 221 (82.2%) | 102 (79.1%) | 255 (85.0%) | 122 (74.8%) | 476 (83.7%) | 224 (76.7%) |
Mobile | 33 (12.3%) | 12 (9.3%) | 28 (9.3%) | 15 (9.2%) | 61 (10.7%) | 27 (9.2%) |
Tablet | 15 (5.6%) | 15 (11.6%) | 17 (5.7%) | 26 (16.0%) | 32 (5.6%) | 41 (14.0%) |
Android | 28 (10.4%) | 11 (8.5%) | 27 (9.0%) | 17 (10.4%) | 55 (9.7%) | 28 (9.6%) |
Chromium | 7 (2.6%) | 1 (0.8%) | 8 (2.7%) | 1 (0.6%) | 15 (2.6%) | 2 (0.7%) |
iOS | 23 (8.6%) | 16 (12.4%) | 22 (7.3%) | 24 (14.7%) | 45 (7.9%) | 40 (13.7%) |
Linux | 1 (0.4%) | 1 (0.8%) | 1 (0.3%) | 1 (0.6%) | 2 (0.4%) | 2 (0.7%) |
Mac | 32 (11.9%) | 37 (28.7%) | 52 (17.3%) | 41 (25.2%) | 84 (14.8%) | 78 (26.7%) |
Ubuntu | 5 (1.9%) | 1 (0.8%) | 1 (0.3%) | 0 (0%) | 6 (1.1%) | 1 (0.3%) |
Windows | 173 (64.3%) | 62 (48.1%) | 189 (63.0%) | 79 (48.5%) | 362 (63.6%) | 141 (48.3%) |
Duration discrimination sensitivity was represented by log values of the slope of each logistic regression curve, where a larger value represents steeper slope and better sensitivity. Participants were randomized to either the seconds or subseconds group and two logistic regression models were fitted for each participant, one for each of the audio and the visual task. Mann–Whitney U tests were used to compare the sensitivity and error between the PD and HC groups for each combination of reference duration (seconds vs. subseconds) and stimulus modality (auditory vs. visual) (
Number of control and patient participants recruited, excluded, and included for analysis. HC, healthy control; PD, Parkinson’s disease patients. Other neurological conditions included multiple sclerosis and epilepsy.
Sensitivity, calculated using equation (
Standard deviations of response times and IIV for each auditory and visual stimuli in the seconds and subseconds range are summarized in
Standard deviation (SD) and intra-individual variability (IIV) of response time in HC and PD groups responding to auditory and visual stimuli in the seconds and subseconds range.
Subseconds |
Seconds |
|||||
HC ( |
PD ( |
HC ( |
PD ( |
|||
Mean (SD) | 447 (310) | 570 (506) | 0.014 | 458 (327) | 519 (283) | 0.049 |
Median (Min, Max) | 350 (105, 2,740) | 405 (115, 4,100) | 364 (106, 2,710) | 448 (148, 1,470) | ||
Mean (SD) | 445 (324) | 631 (918) | 0.033 | 477 (390) | 749 (1,760) | 0.07 |
Median (Min, Max) | 368 (93.3, 2,710) | 440 (85.2, 9,390) | 380 (113, 3,330) | 508 (133, 20,600) | ||
Mean (SD) | 8.54 (0.796) | 8.73 (0.761) | 0.03 | 8.56 (0.834) | 8.65 (0.858) | 0.288 |
Median (Min, Max) | 8.60 (6.16, 9.95) | 8.66 (7.21, 9.95) | 8.60 (6.93, 10.0) | 8.66 (6.71, 10.0) | ||
Mean (SD) | 8.89 (0.762) | 9.06 (0.775) | 0.055 | 8.67 (0.787) | 8.76 (0.814) | 0.259 |
Median (Min, Max) | 9.00 (7.14, 9.95) | 9.14 (7.14, 9.95) | 8.72 (6.86, 10.0) | 8.77 (6.71, 10.0) |
This study has shown that patients with PD have impaired sensitivity in discriminating between durations of both visual and auditory stimuli compared to HCs. This effect has not been found in previous studies (
To the authors’ knowledge, this is the first study of time perception in PD to have been conducted entirely online. While this has enabled us to greatly increase sample sizes, several sources of variability are also introduced by this method of delivery.
Firstly, computer hardware and software varied significantly between participants (see
Secondly, we relied entirely on self-report measures; participants were not assessed by a clinician as part of the study, nor were any data from their medical records available. Patients on antiparkinsonian medications were asked to list their medications and estimate the number of hours since last taking medication. Thus, patients were not in a well-defined “on” or “off” state during testing, and levodopa equivalent dose could not be estimated from this information. While a previous small study found no effect of levodopa on PD patients’ ability to discriminate between durations around 50 ms (
Finally, the experimental environment cannot be controlled as tightly as is typical for in-person experiments. Factors intrinsic to the task such as stimulus luminance/loudness and contrast, which are known to affect duration perception for visual stimuli (
Despite these limitations, online delivery of this study has permitted group sizes an order of magnitude greater than those used in previous studies, and recruitment across a far larger geographical area, leading to greater statistical power.
There is a significant body of evidence to support the theory that time perception is mediated by different mechanisms at different timescales (
Taken together, this model would predict that PD, which is characterized by degeneration of SNc dopaminergic neurons and widespread striatal pathology, produces an impairment in suprasecond interval perception across modalities. However, the nature of this impairment is complex, as one must consider the interacting effects of a hypodopaminergic state, the resulting chronic dysfunction (hyperexcitability) in striatal dopamine-
Biases in temporal perception have also been observed in this study, with systematic differences in the relative perceived durations of the standard (first) and comparison (second) stimuli (
Error of patients and controls in comparing auditory and visual stimuli in the seconds and subseconds intervals (Equation 3). No significant difference was seen between the patient and control groups for the visual stimuli. For the auditory stimuli, PD patients overestimated the standard duration compared to the controls. Annotations denote adjusted
The effect first observed by
Care must be taken when interpreting these effects, as perception may change based on previous stimuli or peri-stimulus sensory or motor events. For example, repetition of a stimulus shortens its perceived duration (
Additionally, fatigue may introduce biases which may differ both between experimental groups and within the patient population. On balance, it is difficult to completely adjust for these in studies such as ours, but further studies should attempt to quantify these effects specifically, with a larger number of participants and possibly a fixed, higher number of trials to elicit fatigue in HCs.
Aberrant time perception in PD could provide some insight into disease progression and severity. Perceptual error in auditory duration comparison correlated with mini-mental state examination (MMSE) score in a recent study (
Data on disease duration and medication were collected as part of our study, and will be included in a future analysis.
Our findings demonstrate modality- and stimulus length-specific differences in duration perception in patients with PD vs. HCs. These differences may shed light on timekeeping systems and the pathophysiology of PD, but require further study and careful examination of the effects observed as they can be sensitive to task design and experimental conditions.
The original contributions presented in this study are included in the article/
The protocol was approved by the University of Oxford Medical Sciences Division Research Ethics Committee (R64730/RE001). The patients/participants provided their written informed consent to participate in this study.
ZS, SP, CA, and JF designed and oversaw the study. ZS and CA oversaw the testing and implementation of the interactive website (
CA receives funding from the NIHR, UCB – Oxford Collaboration. JF was supported by the National Institute for Health Research (NIHR) Oxford Biomedical Research Centre (BRC).
We would like to thank Parkinson’s UK (
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
The Supplementary Material for this article can be found online at: