How does social distancing affects time perception and cognitive load?
TIME PERCEPTION DURING CONFINEMENT
Student : Cyril Nicolaï Internship tutor : Virginie van Wassenhove Date : 11/12/20 Internship presentation Outlines
● Context Time perception and confinement ~ 5 min
● Methods The N-back dual task paradigm ~ 5 min
● Results WM load and confinement effects ~ 5 min Time distortion
La persistance de la mémoire, Salvador Dalí
In 1962, Michel Siffre spent 2 months living in a cave, with no temporal cues: “I had to count from 1 to 120, at the rate of one digit per second. [...] It took me five minutes to count to 120. In other words, I psychologically Caveman: An Interview with Michel Siffre, 2008 experienced five real minutes as though they were two”. Attention and time perception
good or overestimation underestimation of of the event’s duration the event’s duration
Slowing of the passing Speeding of the passing of time judgment of time judgment
bored / passive engaged / active
Attention is not Attention is diverted diverted away from time + cognitive load?
Passing of time judgment ≠ Duration estimation (Droit-Volet, Trahanias,& Maniadakis, 2017) A model of duration estimation
Interval-based Internal Clock arousal ATTENTION (adapted from Treisman, 1963)
+/-
increase pacemaker ACCUMULATOR linearly produces tics of the clock comparison
WORKING LONG TERM MEMORY MEMORY Cognition and Brain Dynamics team
CEA, DRF/Joliot, NeuroSpin, INSERM, Cognitive Neuroimaging Unit (UNICOG)
Team : Cognition & Brain Dynamics Director : Virginie van Wassenhove
METHODS INVESTIGATES
Temporal cognition Cognitive theory & Human neuroscience methods Interval sens Phase sens psychophysics fMRI quantify/compare map events M/EG time duration in time Time social distancing: an ambitious project
N = 353 N = 68 Longitudinal Session 1 Session 3 online Study confined/unconfined during confinement + 90 days after confinement ~ 13 questionnaires Analysing only one (Demographics, task Loneliness scale, ~ 6 questionnaires Anxiety and Depression scale...) French data 3x runs! + 3x runs! + 38 subjects are ~ 10 longer tasks studied for S1 and S3 ~ 10 longer tasks (< 5 min) (< 5 min) + strengths + ~ 3 short tasks → many factors and data ~ 5 short tasks (< 1 min) (< 1 min) → compare data from confinement and no-confinement sessions Time perception during Covid-19 lockdown
Time and Covid-19 stress in the lockdown situation The passage of time during the UK Covid-19 lockdown (Droit-Volet, Gil, et al., 2020) (Ogden, 2020)
❖ Passage of time during ❖ Retrospective passage of time the present moment (across days and weeks)
❖ Distortion: slowing of time passage ❖ Distortion: slowing of time passage but also speeding of time passage
❖ Factors: - Boredom ❖ Factors: - Age - Sadness - Stress - Task load - Social interaction
slowing of time:
● increase with state of sadness and low degree of immersion in an activity (Droit-Volet and Wearden, 2015) ● increase with economic and personal insecurity and suffering (Flaherty, Freidin et Sautu, 2005) Outlines
● Context Time perception and confinement ~ 5 min
● Methods The N-back dual task paradigm ~ 5 min
● Results WM load and confinement effects ~ 5 min N-back Dual Task Paradigm
Task duration Duration Passing estimation of time
1-back ? ? G T V V T “In the last round, 1-back correct the time seemed to response “In the last have passed:” round, how much time has - Very slowly elapsed? - Slowly (mm:ss)” - Normally Task duration - Quickly - Very Quickly
3-back ? ? G T V V T Duration Passing 3-back correct estimation of time response ExperimentExperiment Design design
x3 runs 1-back 3-back 3 runs per Participant per Session 45s 45s 30 ILI = 1500ms ILI = 1500ms letters 2x2x2 design : N = 1 N = 3 8 n-back tasks per Run per Session 90s 90s 60 ILI = 1500ms ILI = 1500ms letters N = 1 N = 3 N-back : N = 1 or N = 3 ⇒ manipulates working memory (WM) load 45s 45s 25 ILI = 1800ms ILI = 1800ms letters Task duration : 45 or 90s N = 1 N = 3 ⇒ manipulates task duration to estimate
90s 90s 50 Interval Inter letter (ILI) : 1500 or 1800ms ILI = 1800ms ILI = 1800ms letters ⇒ manipulates presentation speed N = 1 N = 3 Linear Mixed Effect models
WM load to explain the variance of the Task duration Duration deviation Δt = duration estimate - task duration Inter-letters interval
Subject n°3 each subject made 1 to 24 Subject n°5 estimates for each session: Subject n°1
Subject n°4
Subject n°2 Multi-level model Linear mixed effect models | #2 © mfiz hierarichal model - Michael Freeman
with j the subject and i the estimate
Duration deviation = WM load * Duration * ILI + (1 + Duration + WM load | Subjects)
Fixed effects (+ interactions) Random effects → common to the whole sample → specific to the subjects Outlines
● Context Time perception and confinement ~ 5 min
● Methods The N-back dual task paradigm ~ 5 min
● Results WM load and confinement effects ~ 5 min Effects of attention and WM load on duration estimates
(for prospective dual tasks) single dual task task The effect of attention and working memory on the estimation of elapsed time (Polti, Martin, van Wassenhove, 2018)
Attention: Single task vs Dual task → diverting attention away from time shortened perceived duration
Working Memory load: n=1 vs n=2 vs n=3 → increasing WM load decreased subjective duration and this effect scaled with durations
Fig3.a from Polti, Martin, van Wassenhove, 2018 N-back Performance vs task duration and WM load (S1)
The performance is higher in 1-back tasks
Normal scores and reaction times
The task worked!
Better reaction times but lower correct response rate in 90s tasks DD for different ILI, WM load and durations (S1)
Main effect of task duration = -18.36s*** Interaction ILI | 90s/3-back = 3.19s*** Interaction WM load | 1800ms/90s = - 4.24s*** Longitudinal study - Duration estimation
Main effect session* S1: -17.5s | S3: -13.4s
Main effect WM load* 1-back: -14.5s | 3-back: -16.4s
Main effect duration*** 45s: -5.52s | 90s: -25.44s
duration Higher underestimation of durations during the confinement, scaling with WM load (prospective tasks S1 and S2 → 38 subjects, ~3000 observations) Conclusion and perspectives
The n-back and dual task worked Refine measure of → N-back performance and cognitive engagement 1 systematic underestimation of duration
effects of session and WM Further analysis of the longitudinal study 2 load on duration estimation + Control session
Help researchers from abroad and 3 investigates cross cultural difference Bibliography
Foer, J. (2008). Caveman: An Interview with Michel Siffre. Cabinet Magazine, 30.
Wearden J, O’Donoghue A, Ogden R, Montgomery C. 14 Subjective Duration in the Laboratory and the World Outside. Subjective time: The philosophy, psychology, and neuroscience of temporality. 2014 Apr 4:287.
Droit-Volet, S., Trahanias, P., & Maniadakis, M. (2017). Passage of time judgments in everyday life are not related to duration judgments except for long durations of several minutes. Acta psychologica, 173, 116-121.
Treisman M. Temporal discrimination and the indifference interval: Implications for a model of the" internal clock". Psychological Monographs: General and Applied. 1963;77(13):1.
Droit-Volet S, Sandrine GI, Martinelli N, Andant N, Clinchamps M, Parreira L, Rouffiac K, Dambrun M, Huguet P, Dubuis B, Pereira B. PONE-D-20-12336. Time and Covid-19 stress in the lockdown situation: Time Free, Dying of Boredom and Sadness.2020
Ogden RS. The passage of time during the UK Covid-19 lockdown. Plos one. 2020 Jul 6;15(7):e0235871.
Flaherty, M. G., Freidin, B., & Sautu, R. (2005). Variation in the perceived passage of time: a cross-national study. Social Psychology Quarterly, 68(4), 400-410.
Knoblauch K, Maloney LT. Mixed-Effects Models. In Modeling Psychophysical Data in R 2012 (pp. 257-301). Springer, New York, NY.
Blacker KJ, Negoita S, Ewen JB, Courtney SM. N-back versus complex span working memory training. Journal of cognitive enhancement. 2017 Dec 1;1(4):434-54.
Polti I, Martin B, van Wassenhove V. The effect of attention and working memory on the estimation of elapsed time. Scientific reports. 2018 Apr 27;8(1):1-1.
Juvina I, Taatgen NA. Modeling control strategies in the n-back task. InProceedings of the 8th international conference on cognitive modeling 2007 Jul (pp. 73-78). Psychology Press New York. NY.
IMAGES : All images are credited on the slides or are from me: © Cyril Nicolaï
créditer image site all other diagram, draws or graphs are mine N-back dual task : example
Sample trial from Polti, Martin, van Wassenhove (2018). Scientific reports, 8. Youtube.
Attention, no feedback in the TSD n-back! Longitudinal study - Passing of time judgment
● Interaction between WM load and sessions? Session and WM load ● No statistical analysis for the moment Measures
nb of Hits + nb of Correct Rejections Correct Response Rate = nb of letters
Duration deviation = subjective duration estimate - task objective duration
subjective duration estimate - task objective duration Normalised = Duration Deviation task objective duration
Possible responses of the participants Linear mixed effect models
fig: LME models assessing effect of different conditions on duration deviation Δt (Session 1, Run 1)
you are the chosen one!
Duration deviation = WM load * Duration * ILI + (1 + Duration + WM load | Subjects)
Fixed effects (+ interactions) Random effects → common to the whole sample → specific to the subjects Distribution of duration deviation
● rounded up
● verbalisation effect
● Scalar effect of duration
Duration deviation = subjective duration estimate - task objective duration
Scalar effect of duration
Scalar variability normalise the duration estimations N-back Performance vs ILI and WM load
No much insight on the ILI in the literature
Subjects Driven by the pace of the presentation rate? → higher false alarm rate
confounded by a strong learning effect? (Juvina & Taatgen, 2007)
(prospective tasks S1) N-back Performance across sessions
● ceiling effect
● training effect
● better performance induces...
(prospective tasks from S1 and S3) N-back task across runs (S1)
● Significant differences between duration estimates from Run 1 and those from Run 2 or Run 3 *** ● Performance increases with the runs ● No significant difference between Run 2 and Run 3 (p = 0.33) ● training effect (with ceiling effect) ● Training effects + tasks no longer retrospective ?