VR based Collaborative Errorless Learning System using Humanoid

Avatar for People with Alzheimer’s Disease

Takehiko Yamaguchi

, Yuta Yamagami

, Toshihiko Sato

, Hibiki Fujino

, Tetsuya Harada

Debora Alexandora Foloppe

3,4

, Paul Richard

and Philippe Allain

3,5

The Department of Appliced Electronics, Tokyo University of Science, Niijuku6-3-1, Katsushikaku, Tokyo, Japan

Faculty of Industrial Science and Technology Applied Electronics, Tokyo University of Science, Niijuku6-3-1,

Katsushikaku, Tokyo, Japan

Laboratoire de Psychologie des Pays de Loire (UPRES EA 4638), Université d’Angers, Angers, France

Laboratoire Angevin de Recherche en Ingénierie des Systèmes (EA 7315), Université d’Angers, Angers, France

Unité de Neurologie, Département de Neurologie, CHU d’Angers, Angers, France

Keywords: Errorless Learning, Alzheimer’s Disease, Virtual Agent, Collaborative Cognition, Mind-Wandering.

Abstract: Everyday action impairment is one of the diagnostic criteria of Alzheimer’s disease and is associated with

many serious consequences, including loss of functional autonomy and independence. It has been shown

that the (re)learning of everyday activities is possible in Alzheimer’s disease by using errorless learning

approaches. The purpose of this study is to propose a newly revised Virtual Kitchen system that allows

training of everyday activities to integrate a new approach of errorless learning (EL) framework using

collaborative learning with a virtual agent. In this paper, we describe a concept of the proposed framework,

as well as explore user’s attention change to analyse eye tracking data during a training task in order to

review the effectiveness of the proposed EL framework.

1 INTRODUCTION

1.1 Everyday Activities Intervention

Everyday activities are familiar tasks that require

multiple cognitive processes such as serial ordering

of task steps and object selection, in order to achieve

practical goals such as preparing a cup of coffee,

making a sandwich, and so on. Usually these

familiar tasks are routinely performed with

subjective ease in real life. However, in people with

brain damage, errors are frequent and sometimes

may preclude the achievement of the task goal

(Chevignard et al., 2008; Fortin et al., 2003). In fact,

everyday action impairment is one of the diagnostic

criteria of Alzheimer’s disease as well as is

associated with many serious consequences such as

institutionalization, depression, and death (Adam et

al., 2000; Hargrave et al., 2000; Noale et al., 2003).

In a common approach to prevent these dire

consequences, neuropsychologists suggest that

patients with dementia learn or relearn instrumental

activities of daily living to increase the patients’

functional autonomy (Giovannetti et al., 2007;

Dechamps et al., 2011). To date, research on

everyday action interventions for people with

dementia has shown that repetition of everyday

activities improves performance on trained tasks

(Avila et al., 2004; Farina et al., 2002).

1.2 VR based Intervention

Today, there are no doubts that (re)learning methods

have beneﬁcial effects for patients with Alzheimer’s

disease especially when they have explicit memory

deﬁcits. However, conventional methods are not

always feasible in typical clinical settings.

Virtual reality (VR) based technology is one of

the emerging tools that has great potential for use in

rehabilitation (Le Gall et al., 2008). VR is a

technology that allows people to view, touch,

navigate, as well as interact with a computer-

generated three-dimensional world in real time. In

addition, Virtual environments allow people to be

engaged in activities as in the real world (Broeren et

al., 2008). In the domain of everyday activities,

several virtual environments have been developed to

462

Yamaguchi T., Yamagami Y., Sato T., Fujino H., Harada T., Alexandora Foloppe D., Richard P. and Allain P..

VR based Collaborative Errorless Learning System using Humanoid Avatar for People with Alzheimer’s Disease.

DOI: 10.5220/0005313904620469

In Proceedings of the 10th International Conference on Computer Graphics Theory and Applications (GRAPP-2015), pages 462-469

ISBN: 978-989-758-087-1

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

simulate daily tasks, namely cooking (Allain et al.,

2011; Richard et al., 2010; Zhang et al., 2002) as

well as shopping (Klinger et al, 2006). In fact, VR

allows intensive repetition of meaningful tasks with

augmented feedback for rehabilitation (Crosbie et

al., 2012).

Since no study so far has captured the scope to

create a virtual environment speciﬁcally designed

for the (re)learning of everyday activities in patients

with Alzheimer’s disease, we have designed a dual

modal VR application which allows patients with

Alzheimer’s disease to practice everyday activities:

Virtual Kitchen (VK) system (Yamaguchi et al.,

2012). In the previous study, we have developed the

VK for use as an intervention tool to treat everyday

cooking deﬁcits in patients with Alzheimer’s

disease, as well as examined the effectiveness and

usability.

1.3 Virtual Kitchen System

Previous VK system was designed as a dual modal

VR platform. The system was composed of one

notebook PC, a mouse, a headset, and the VK

application, which runs on the PC. The system was

designed to be as simple as possible in order to make

it portable and to facilitate setup. The application

was visually implemented in a 3D environment in

order to make it immersive as well as improve the

reality of the system. As for the interaction

technique with the system, 2D mouse interaction

was supported for patients since it was the most

common and usual interaction model for them and

since the system focuses on cognitive performance

but not on motor abilities, we decided to control the

vertical and horizontal position of objects in the 3D

environment.

The VK application consisted of (re)learning

everyday tasks, especially cooking tasks such as

preparing two pieces of toast for breakfast using an

electric toaster (breakfast task); and preparing a cup

of coffee with a coffee machine (coffee task). The

breakfast task trains subjects how to prepare two

pieces of toast in everyday life. It contains seven

manipulable visual objects: a toaster, two pieces of

bread, two spoons of jam, and two pats of butter. As

for the coffee task, it trains subjects how to prepare a

cup of coffee with a coffee machine in everyday life.

The coffee task also contains seven manipulable

visual objects: a coffeemaker, a coffee ﬁlter, ground

coffee, a coffeepot, a water pot, sugar cubes, and a

milk carton. In both tasks, these seven objects can be

manipulated with the mouse. To finish both tasks, a

total of 10 manipulations were severally required.

1.4 (Re)Learning Methods

In the previous VK system, we focused on

developing two cooking tasks as descried above on a

dual modal VR platform with integration of two

(Re)learning methods: Written Instruction Learning

Method (Written ILM), and Self-Recorded

Instruction Learning Method (Self-Recorded ILM).

Both methods provide step-by-step written

instructions to achieve the applied task. These

instructions were proposed to reduce errors during

(re)learning. Written ILM was based only on written

instructions, while the Self-Recorded ILM combined

written ILM with verbal repetition of instructions.

In fact, written and/or verbal instructions

approaches have proven to be successful in patients

with action disorganization problems (Bickerton et

al., 2006). Such approaches could be very helpful to

guide their actions. However, this approach could

only provide errorful learning which is not

associated with an avoidance of errors during

leaning.

1.5 Errorless Learning

Errorless learning (EL) is one of the effective

learning methods for everyday activities intervention.

The EL allows teaching techniques that prevent

people from making mistakes during learning

process (Dechamps et al., 2011).

Some recent studies suggest that errorless learning

principles may be beneficial in memory

rehabilitation for people with dementia (Dunn &

Clare, 2007). In addition, EL has been applied in

developing practical interventions for people with a

range of disabilities so that recently EL has been an

important focus of interest for researchers interested

in memory processes (Clare & Jones, 2008).

1.6 VR based Errorless Learning

Although there are many evidences based studies

and successful cases that support the use of the EL,

usually EL approaches are conducted by therapists,

but not by system operation. On reviewing the

literature, we found a few studies on the EL

approach implemented in VR environments. For

instance, Connor et al tested the effectiveness of

their errorless learning system using haptic guidance

with patients with post-stroke visuo-perceptual

deficits. In their errorless learning condition

unproductive or incorrect approaches to objects

within a virtual environment are prevented by

applying a ‘force field valley’ when the patient take

VRbasedCollaborativeErrorlessLearningSystemusingHumanoidAvatarforPeoplewithAlzheimer'sDisease

463

a wrong direction (Connor et al., 2002). Kober et al.

developed VR based neurologic rehabilitation

system for patient with spatial disorientation. Their

EL system was designed by means of verbal

feedback from the experimenter when patients took

a wrong route during navigating in virtual scenarios

(Kober et al., 2013). At least according to the related

studies, researchers focus on how well instruction

provide from the system helps a patient better

understand his or her performance, as well as how

patient’s error behaviour can be prevented by using

multimodal VR interaction when they make an error

during rehabilitation task.

In fact, typical algorithm of the EL usually

consists of a couple of processes: “(a) breaking

down the target task into small to discrete steps or

units; (b) providing sufficient models or instruction

before a patient is asked to perform the target task;

immediately correcting errors, and (e) carefully

fading prompts” (Sohlberg et al., 2005). As the point

of view of the system requirement when designing

an EL framework, it should be considered that not

only the typical process of the EL, but also (1) how

well patient’s attention can be induced to the

instruction and the target task, (2) how naturally

patient’s motivation can be created for them to

conduct the target task, (3) how naturally patient’s

error behaviour can be predicted, as well as (4) how

effectively prompts can be reduced.

In this paper, we propose a newly revised VK

system to integrate an EL framework for (re)learning

of everyday activities in patients with Alzheimer’s

disease. The proposed EL framework was designed

to achieve these requirements as described above.

We present an implemented prototype system as

well as the effectiveness of the system’s capability.

2 PROPOSED SYSTEM

2.1 Requirements Analysis

The purpose of the requirements analysis process is

to define the functionalities needed in the proposed

EL framework. Four main requirements of the

proposed framework were actually extracted based

on the discussion in the previous section: (1) control

a visual attention of a patient, (2) create a natural

situation in which a patient can be motivated into an

applied task, (3) predict a patient’s error behaviour,

as well as (4) adaptively provide a level of prompt

based on patient’s progress. To achieve these

requirements, we decided to develop an EL

framework based on a collaborative learning

environment with a humanoid avatar which is a

virtual agent controlled by an artificial intelligence

(AI) technique.

Figure 1: System framework of the proposed VK system.

Humanoid agent has a capability of supporting

learning activities through improving learner’s

emotional state such as engagement as well as

motivation (Soliman & Guetl, 2011). Using

collaborative learning with the avatar, requirements

(1), (2) can be achieved. In terms of requirement (3),

error behaviour detection framework is considered.

Concerning requirement (4), dynamic difficulty

level adaptation framework is considered (Figure 1).

In addition, we revised the interaction technique

with the proposed VK platform using Leap motion

device, which is a sensor device that supports hand

and finger motion as input, to facilitate natural

communication with the humanoid avatar.

2.2 System Framework

The prototype system is composed of a desktop PC,

a PC monitor and Leap motion controller (Figure 2).

In this paper, four main frameworks are described in

detail such as a Task management framework, an

Interaction technique management framework, an

Avatar control framework, and an Error behaviour

detection framework.

2.2.1 Interaction Technique Management

Framework

This framework enables to control a hand avatar

which tracks the user’s hand gesture using the

connected Leap motion controller. The position and

orientation data of the user’s hand are updated at

about 125Hz. However, since the data confidence of

the Leap motion becomes low due to a tracking error

and self-occlusion, the orientation data of the user’s

hand avatar is fixed while grabbing some

GRAPP2015-InternationalConferenceonComputerGraphicsTheoryandApplications

464

manipulable object to improve the usability of the

hand avatar.

Figure 2: System configuration of the VK system.

2.2.2 Task Management Framework

This framework currently enables to manage a

coffee making task. The screen shot of the coffee

making task is shown in the Figure 3. The coffee

making task contains seven manipulable visual

objects such as a coffeemaker, a coffee ﬁlter, ground

coffee, a coffeepot, a water pot, sugar cubes, and a

milk carton. These seven objects can be manipulated

with the hand avatar. To finish the tasks, a total of

10 manipulations were required as shown below.

1. Press the blue button on the coffee maker to

open the filter door.

2. Set a coffee filter in the filter door.

3. Pour the ground coffee into the filter door.

4. Press the blue button again to close the filter

door.

5. Pour the water from water pot into the coffee

maker.

6. Set the coffee pot on the warming plate.

7. Press the red button on the coffee maker to

make coffee.

8. Pour the coffee from the coffee pot into the

coffee cup.

9. Put a cube sugar from the sugar plate into the

coffee cup.

10. Pour the milk from the milk carton into the

coffee cup.

2.2.3 Avatar Control Framework

The humanoid avatar in the VK platform is

controlled by AI technique which is running on the

Avatar control framework. Figure 4 shows that

information flow between AI and the environment.

AI is running on the background of the main task. AI

has a sensor module which allows obtaining

environmental parameters such as a current task

status, task success/failure, distances to each object,

as well as position/orientation of the user’s hand

avatar.

Figure 3: Screenshot of the coffee making task.

The AI control module consists of three components

such as a Knowledge making component, a Decision

making component, and a Motion making

component. Obtained information through the sensor

module is recognized in the Knowledge making

component to analyse patient’s behaviour as well as

current task status. The result from the Knowledge

making component is updated in the Database to

share it with other components.

Figure 4: Information flow between AI and the virtual

environment.

Depending on the result from the Knowledge

making component, next operation of the avatar is

decided by the Decision making component as well

as by the Motion making component to update an

operation mode of the avatar which consists of three

operation modes: a Collaborative operation mode, an

Inductive operation mode, and a Leading operation

mode. In fact, these operation modes could be a key

concept for designing collaborative learning

environment.

The Inductive operation mode is a set of avatar’s

behaviour that allows indicating the next task for a

patient to achieve. When a task begins, the avatar

VRbasedCollaborativeErrorlessLearningSystemusingHumanoidAvatarforPeoplewithAlzheimer'sDisease

465

immediately moves to a target object. For example,

if the applied task is the trial 2 (See 2.2.2 section),

the avatar moves to the coffee filter and then

performs an inductive behaviour (the avatar jumps

up and down, or shakes the hand) to encourage a

patient to conduct the target task. When it succeeds

to indicate patient’s attention to the target task, the

avatar’s behaviour is changed to the Collaborative

operation mode.

Figure 5: Collaborative operation of the avatar.

When the avatar is in the Collaborative operation

mode, first, the avatar tries to achieve a target task

alone (Figure 5). For example, if the target task is

trial 2, the avatar tries to push the filter, however it

does not move after all. In this situation, the avatar is

asking for somebody’s help. This situation actually

was designed to encourage patient’s helping

behaviour to the avatar. With this approach, we

designed a natural situation in which a patient to be

motivated into an applied task. If it fails to indicate

patient’s attention to the target task, the Leading

operation mode is applied to demonstrate a model

prompts of the target task by the avatar.

2.2.4 Error Behaviour Detection

Framework

This framework is independently working as part of

the sensor module in the Avatar control framework.

This framework enables to analyse hand avatar’s

trajectory as well as velocity distribution to predict

error behaviour of a patient. According to the motor

control theory, arm movement can be estimated

using minimum Jerk model. We simply integrated

this model to estimate avatar’s trajectory.

3 SYSTEM EVALUATION

The goal of this evaluation was to explore how

different eye movement patterns can inform us about

cognitive activities between subjects with and

without the avatar communication during the coffee

making task. The following section describes the

selected participants, the materials, the procedure

and the data analysis of this evaluation.

3.1 Subjects

15 men and 1 women aged 21 to 25 were recruited

through the Tokyo University of Science to

participate in this study (age M = 22; SD = 1.20).

All subjects were used to handle computers;

however none of them had experience playing video

games.

3.2 Material

The experimental setup consisted of a PC (Intel Core

i7-4770 CPU with 3.40 GHz) with a 21.5" BenQ

screen (1920x1080), and a leap motion controller.

The virtual environment was developed using the

system Unity 4.5 which was the game development

platform. Tobii X2-30 eye tracking system was used

to measure their eye movement patterns during the

experiment. The eye tracker was attached under the

display.

3.3 Procedure

Each participant tried out all two coffee making

tasks with/without the avatar communication in a

counterbalanced order. The subjects were seated in

front of a screen monitor and first received general

verbal information about the evaluation, the learning

method, and the use of the VK system.

The subjects received two pre-training sessions

to familiarize them with the method (i.e., with the

avatar or without the avatar). They were in the same

condition as the learning condition. The pre-training

sessions were designed to familiarize them with the

use of the Leap motion controller and the visual

instructions. After these two pre-training sessions,

the subjects were given a learning session, during

which a task was performed using with the avatar or

without the avatar. During the learning session, their

eye movements were measured by using the eye

tracking system.

GRAPP2015-InternationalConferenceonComputerGraphicsTheoryandApplications

466

4 RESULTS AND DISCUSSION

Eye movement pattern during the virtual coffee

making task with/without the avatar was measured.

Figure 6 shows that the example of the observed eye

movements in the case of the task with the avatar.

The dashed line box represents an area that is set

to count how often a fixation point were appeared to

calculate a state transition rate of the fixation point.

These areas were focused on the manipulable objects,

instruction area, the avatar, as well as avatar’s

message which is represented on the avatar.

The average state transition rate was calculated

using the obtained eye movement data during the

conducted experiment. However, the data of the

three subjects out of the 16 subjects were removed

from the analysis since there was missing value on

their raw data.

Table 1 shows that the average state transition

rate of the fixation point in the coffee making task

(in without avatar condition). Where, the others state

in the table is a transition rate counted when the

subject’s fixation point is out of the focused area in

Figure 6.xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

The result shows that the most frequent transition in

this case was 30.89 %, namely the transition from

the Target object state to the others state. In

opposite transition direction the result was also high:

28.54%.xxx

Figure 6: Example of the observed eye movements.

In fact, when the task begins, first of all, a subject

paid attention to the instruction on the screen. Then

if the subject correctly understands the applied task,

he/she may take a look at the target object. From the

results, however, the subjects mostly see the other

objects (11.86%) after the instruction disappeared

from the screen. Table 2 shows that the average state

transition rate of the fixation point in the coffee

making task with avatar. The result shows that the

most frequent transition in this case was 10.09 %,

namely the transition from the Avatar state to the

Target object state. The transition from the Target

object state to the Avatar state is also high: 10.02%.

After the instruction on the screen disappeared,

most of subjects paid attention to the others as same

as without avatar condition. However, the transition

rate of the Target objects state, and Avatar state are

also respectively high.

Table 1: Average state transition rate (%) of the fixation

point in the coffee making task without avatar.

Table 2: Average state transition rate (%) of the fixation

point in the coffee making task with avatar.

According to some subject’s comments, he/she felt

his/her mind-wandering during the experiment in

without avatar condition. The mind-wandering is a

state of decoupled attention. In this state, usually

task-unrelated thoughts also occur.

In contrast, with avatar condition, most of

subject reported that they felt friendly toward the

avatar, as well as the avatar encouraged a helping

behaviour.

5 CONCLUSIONS

We proposed a newly revised Virtual Kitchen

system to integrate a new approach of errorless

learning (EL) framework using collaborative

learning with a virtual agent. In this paper, we

described a concept of the proposed system, as well

as conducted an evaluation to explore the

effectiveness of the proposed EL framework. As the

result of the evaluation using eye movement pattern

analysis, we found that subject’s visual attention was

successfully induced by the avatar. In addition,

according to the subjective report, we found that the

possibility to prevent user’s mind-wandering using

the proposed framework.

VRbasedCollaborativeErrorlessLearningSystemusingHumanoidAvatarforPeoplewithAlzheimer'sDisease

467

ACKNOWLEDGEMENTS

This work was financially supported in part by a

Grant-in-Aid for Young Scientists (B) from Japan

Society for the Promotion of Science.

REFERENCES

Adam, S., Van der Linden, M., Juillerat, A.C., & Salmon,

E., 2000. The cognitive management of daily life

activities in patients with mild to moderate

Alzheimeer’s disease in a day care center: A case

report. Neuropsychological Rehabilitation, 10(5), 485-

509.

Allain, P., Richard, P., Naud, M., Besnard, J., Massenot,

L., Nolin, P., & Le Gall, D., 2011. The value of a

virtual kitchen to assess the activities of daily life in

Alzheimer’s disease. Journal of Cybertherapy and

Rehabilitation, 4(2), 278–279.

Avila, R., Bottino, C., Carvalho, I., Santos, C., Seral, C.,

& Miotto, E., 2004. Neuropsychological rehabilitation

of memory deficits and activities of daily living in

patients with Alzhheimer’s disease: A pilot study.

Brazilian Journal of Medical and Biological Research,

37(11), 1721-1729.

Bickerton, W. L., Humphreys, G. W., & Riddoch, J. M. ,

2006. The use of memorised verbal scripts in the

rehabilitation of action disorganisation syndrome.

Neuropsychological Rehabilitation, 16(2), 155–177.

Broeren, J., Claesson, L., Goude, D., Rydmark, M., &

Sunnerhagen, K. S., 2008. Virtual rehabilitation in an

activity centre for community-dwelling persons with

stroke: The possibilities of 3-dimensional computer

games. Cerebrovascular Disorders, 26(3), 289–296.

Chevignard, M., Taillefer, C., Picq, C., Poncet, F.,

Noulhiane, M., & Pradat-Diehl, P., 2008. Ecological

assessment of the dysexecutive syndrome using

execution of a cooking task. Neuropsychological

Rehabilitation, 18(4), 461-485.

Clare, L., Jones, RS., 2008. Errorless learning in the

rehabilitation of memory impairment: a critical review.

Neuropsychol Rev, 18(1), 1-23.

Connor, B. B., Wing, A. M., Humphreys, G. W.,

Bracewell, R. M., and Harvey, D. A., 2002. Errorless

learning using haptic guidance: Research in

cognitive rehabilitation following stroke. In

Proceedings of the. 4

International Conference on

Disability, Virtual Reality & Associated Technology,

Veszprém, Hungary, 77-84.

Crosbie, J. H., Lennon, S., Mc Goldrick, M. C., McNeill,

M. D. J., & McDonough, S. M. “Virtual reality in the

rehabilitation of the arm after hemiplegic stroke: a

randomized controlled pilot study,” Clinical

Rehabilitation, vol. 26, no. 9, pp. 798–806, 2012.

Dechamps, A., Luciano, F., Jungheim, J., Leone, E.,

Dood, E., Allioux, A., Robert, P. H., Gervais, X.,

Maubourguet, N., Olde Rikkert, M. G. M., & Kessels,

R. P. C., 2011. Effects of different learning methods

for instrumental activities of daily living in patients

with Alzhheimer’s dementia: A pilot study. American

Journal of Alzhheimer’s Disease and Other

Dementias, 26(4), 273-281.

Dunn, J., Clare, L., 2007. Learning face-name associations

in early-stage dementia: comparing the effects of

errorless learning and effortful processing.

Neuropsychol Rehabil. 17(6), 735-54.

Farina, E., Fioravanti, R., Chiavari, L., Imbornone, E.,

Alberoni, M., Pomati, S., Pinardi, G., Pignatti, R., &

Mariani, C., 2002. Comparing two programs of

cognitive training in Alzheimer’s disease: A pilot

study. Acta Neurologica Scandinavica, 105(5), 365-

371.

Fortin, S., Godbout, L., & Braun, C. M. J., 2003. A test of

Shallice’s and Grafman’s neuropsychological models

of executive functions with head trauma patients

performing activities of daily living. Cortex, 39(2),

273-291.

Giovannetti, T., Schwartz, M. F., & Buxbaum, L. J., 2007.

The coffee challenge: A new method for the study of

everyday action errors. Journal of Clinical and

Experimental Neuropsychology, 29(7), 690-705.

Hargrave, R., Reed, B., & Mungus, D., 2000. Depressive

symptoms and functional ability in dementia. Journal

of Geriatric Psychiatry and Neurology, 13, 72-77.

Klinger, E., Chemin, I., Lebreton, S., & Marie´, R. M.,

2006. Virtual action planning in Parkinson’s disease:

A control study. Cyberpsychology and Behavior, 9(3),

342–347.

Klinger, E., Marie, R. M., Lebreton, S., Weiss, P. L., Hof,

E., & Josman, N., 2008. The VAP-S: A virtual

supermarket for the assessment of metacognitive

functioning. VRIC ’08, p. 57. Laval, France.

Kober SE, Wood G, Hofer D, Kreuzig W, Kiefer M,

Neuper C., 2013. Virtual reality in neurologic

rehabilitation of spatial disorientation. Journal of

NeuroEngineering and Rehabilitation, 10 (17).

Le Gall, D., Bersnard, J., Louisy, T., Richard, P., & Allain,

P., 2008. Utilisation de la re´alite´ virtuelle en

neuropsychologie. Neuropsy News, 7(4), 152–155.

Noale, M., Maggi, S., Minicuci, N., Marzari, C., Destro,

C., Farchi, G., Scafato, E., Baldereschi, M., DiCarlo,

A., & Crepaldi, G., 2003. Dementia and disability:

Impact on mortality. Dementia & Geriatric Cognitive

Disorders, 16(1), 7-14.

Richard, P., Massenot, L., Besnard, J., Richard, E., Le

Gall, D., & Allain, P., 2010. A virtual kitchen to assess

the activities of daily life in Alzheimer’s disease.

Proceedings of the International Conference on

Computer Graphics Theory and Applications, 378–

383.

Sohlberg, M. M., Ehlhardt, L., & Kennedy, M., 2005.

Instructional techniques in cognitive rehabilitation: A

preliminary report. Seminars in Speech Language

Pathology, 26, 268 – 279.

Soliman, M., Guetl, C., 2011. Evaluation of intelligent

agent frameworks for human learning. 14th.

GRAPP2015-InternationalConferenceonComputerGraphicsTheoryandApplications

468

International Conference on Interactive Collaborative

Learning (ICL), 191-194.

Werner, P., Rabinowitz, S., Klinger, E., Korczyn, A. D., &

Josman, N., 2009. Use of the virtual action planning

supermarket for the diagnosis of mild cognitive

impairment. Dementia and Geriatric Cognitive

Disorders, 27(4), 301–309.

Yamaguchi, T., Foloppe, D, A., Richard, P., Richard, E.,

Allain, P., 2012. A Dual-Modal Virtual Reality

Kitchen for (Re)Learning of Everyday Cooking

Activities in Alzheimer’s Disease. Presence, 21(1),

43-57.

Zhang, L., Abreu, B. C., Seale, G. S., Masel, B.,

Christiansen, C. H., & Ottenbacher, K. J., 2002. A

virtual reality environment for evaluation of a daily

living skill in brain injury rehabilitation: Reliability

and validity. Archives of Physical Medicine and

Rehabilitation, 84(8), 1118–1124.

VRbasedCollaborativeErrorlessLearningSystemusingHumanoidAvatarforPeoplewithAlzheimer'sDisease

469