Content uploaded by Allan Armougum
Author content
All content in this area was uploaded by Allan Armougum on Apr 18, 2020
Content may be subject to copyright.
Cognitive Load Theory (CLT)
✓CLT –ratio of task demand to available resources (Sweller, 2010; Sweller et al.,
1998; Van Merriënboer & Sweller, 2005)
✓Intrinsic load (Sweller, 1994), which is basically related to the format of the task
itself
✓Extraneous load (Sweller & Chandler, 1991), involved in the processing of
information, irrelevant to the learning task
✓Germane load (Sweller et al., 1998), which refers to the ultimate function of
learning, involving schemas creation and automatization of knowledge
(Chanquoy, Tricot & Sweller, 2007; Sweller, Ayres, Kalyuga & Chandler, 2003).
Material: Virtual vs real train station
✓Expertisei= -0.60 + .077*Investigation
train linei+ .079*Investigation train
stationi+ .084*Train stations of the
same train operatori+.078
✓Expertise prediction F(3,120) = 1666,
p< .001, and account for 97.6%
Expertise
✓Long-term working memory (LTWM) framework of Ericsson & Kintsch (1995)
✓Existence of a continuous connection between LTM and WM, where resources
from LTM are allocated to WM, based on the material being processed and its
potential link with stored knowledge.
✓Template Theory of Gobet & Simon (1996a)
✓In experts, enhanced activation of LTM - through different routes, based on a
common Template between the different access routes
✓Example of expert’s memory: Master-level chess player, showing higher
processing activities, storage of 10Kto 100Kgroups of semantically linked
information (Chunks) (Simon and Gilmartin, 1973)
✓Expert reversal effect (Kalyuga, 2009; Kalyuga, Ayres, Chandler & Sweller,
2003), defines a situation in which experts underperform in non-optimal context,
where they can express at most equal performance as novices
Allan Armougum, Chantal Joie –La Marle & Pascale Piolino
Virtual Reality (VR)
✓Allows realistic simulation of daily life behaviors in healthy adults (e.g. virtual
visit of a museum, Carrozzino & Bergamasco, 2010;episodic memory
influenced by action, Plancher, Barra, Orriols & Piolino, 2013; working memory
and episodic memory, Plancher, Gyselinck & Piolino, 2018)
✓Important criteria with VR technology: general experience, spatial presence,
personal involvement and relevant realism (Schubert, Friedmann &
Regenbrecht, 2001), with bodily immersion and sensation (Murray & Gordon,
2001)
✓Important modeling technique: interface, immersion and interaction (Fuchs &
Moreau, 2006)
Hypotheses
✓We expected novice participants to express higher cognitive load than expert
participants in normal context
✓An expertise reversal effect is expected to be observed in experts in non-
optimal context
✓Since environmental context has an impact on cognitive load level, we expect
novice to be slightly disturbed by non-optimal context
INTRODUCTION
EXPERTISE REVERSAL EFFECT: COST OF GENERATING NEW SCHEMAS
Université de Paris, MC2Lab, F-92100
Boulogne-Billancourt, France
allan.armougum@parisdescartes.fr
METHOD
✓Expertise effect, in train travelers (experts vs novices
travelers of the assessed train station)
Expertise
Factors
✓Travelers’profile (expert vs novice)
✓Context (normal vs disturb)
RESULTS
0
5
10
15
20
25
30
35
40
45
50
Factual
(What)
Temporal
(When)
Spatial
(Where)
Total
Binding
Binding score
Memory dimensions
Binding memory score
(behavioral performance analysis)
Expert_normal
Novice_normal
Expert_disturb
Novice_disturb
0
20
40
60
80
100
120
MD PD TD Perf. Effort Frus.
Adjusted load rating
Mental load dimensions
NASA-TLX
(subjective measure)
CONCLUSION & DISCUSSION
✓No profile effect, F(3,120) = .42,p= .89,ƞ² = .025
✓No Context effect, F(3,120) = .23,p= .98,ƞ² = .015
✓No significant interaction, expertise x context effects,
✓F(3,120) = .16,p= .99,ƞ² = .010
✓Saint-Michel Notre Dame train station from Paris, was chosen
(adapted from Armougum, Orriols, Gaston-Bellegarde, Joie-
La Marle & Piolino, 2019)
Navigation path
✓Expertise reversal effect
Experts perform better than novices in normal condition, but reach lower or equal perform
as novices in disturb context
In complex context →Schemas are updated and become enriched and stable (Pollock, Chandler &
Sweller, 2002)
This is in line with WM update (Craik & Lockhart, 1972)→high degree of semantic and cognitive
analysis (Piolino, Desgranges & Eustache, 2009)
Adding new information implies new autobiographical and contextual events: episodic memory
(Tulving, 1972)
New schemas are created from activation of semantic memory →creation of new episodic memory
→conversion to semantic memory
Immersion level
✓General presence,
Spatial presence,
Involvement,
Experienced realism
✓Normal x Disturb
F(3,120) = .40, p = .81,ƞ= .009
✓Expert x Novice
F(3,120) = .85, p = .50, ƞ = .013
✓Virtual reality
VR technique proved to be a potent measuring technique for variations in CL, with
statistically same immersive level in different contexts and expertise levels
✓Expert memory model and Cognitive Load Theory
The insights of expert memory models allowed us to further understand the
generation of schemas in experts, suggesting an upgrade of CLT to better
understand schemas creation in experts, especially in complex context
✓Binding memory score
This measure of performance proves to be a potent measure of behavioral online
aspect of cognitive load
✓Normal x Disturb F(3,120) = 41.0, p < .001, ƞ² = .58
✓Expert x Novice F(3,120) = 4.94, p < .01,ƞ² = .14
✓Normal x Disturb F(3,120) = 1977, p < .001
✓Expert x Novice F(3,120) = 22.8, p < .001
✓Normal x Disturb F(3,120) = 363, p < .001
✓Expert x Novice F(3,120) = 54.1, p < . 001
0
1
2
3
4
5
6
7
8
9
Baseline Phase 1 Phase 2 Phase 3 Phase 4
Mean SCR(CDA)/μs
Train station environment region
EDA - SCR score
(Physiological measure)
Novice_normal
Expert_disturb
Expert_normal
Novice_disturb
