The Biocybernetic Loop Engine: An Integrated Tool for Creating

Physiologically Adaptive Videogames

J. E. Muñoz , E. R. Gouveia, M. S. Cameirão and S. Bermudez I. Badia

Madeira Interactive Technologies Institute, Funchal, Portugal

Universidade da Madeira, Funchal, Portugal

Keywords: Biocybernetic Loop, Videogames, Adaptation, Software, Visual Scripting, Exergames, Heart Rate,

Unity3D.

Abstract: Biocybernetic loops (BLs) are physiological adaptation mechanisms created to augment human-computer

interaction by interpreting human behaviour via physiological responses. Because of its inherent

complexity, the development of BLs has been mainly utilized within the academic environment, with

limited use of physiologically adaptive systems in promising fields such as assistive and gaming

technologies. The Biocybernetic Loop Engine (BL Engine) is an integrated software tool designed for an

easy creation of physiologically modulated videogames by means of wearable sensors. The BL Engine

includes a signal acquisition panel, which facilitates the connectivity of multiple physiological sensors and

the processing of their signals, a biocybernetic console to rapidly create and iterate adaptive rules using a

visual scripting module, and a game connector tool that ties physiological modulations to game variables. In

this paper, we present the BL Engine software architecture, its design and implementation process, as well

as a proof-of-concept of the system applied to an exergaming experience aiming to improve

cardiorespiratory fitness training in older adults. By developing integrated tools that aid the design and

implementation of BLs in videogames, we aim to contribute to the dissemination and widespread use of this

approach in the gaming industry and serious gaming applications.

1 INTRODUCTION

Physiological computing systems are designed to

capture responses of the central and peripheral

nervous systems (Fairclough, 2009). This approach

offers a novel input control between users and

machines (Fairclough and Gilleade, 2014). One use

of such approach is to dynamically adjust systems to

challenge or provide assistance to users (Gilleade et

al., 2005). The concept of physiologically adaptive

systems has been widely developed and documented

following the biocybernetic loop (BL) construct. BL

utilizes the close-loop control, data analysis,

decision making and artificial intelligence from

Wiener’s cybernetics and applies them to

physiological computing (Novikov, 2016). This

method has been used for instance to assist pilots by

detecting their workload levels (Pope et al., 1995);

deliver in an autonomous, timely, consistent and

accurate way therapy/drugs to patients (Loeb and

Cannesson, 2017) (Mishra and Gazzaley, 2014);

adapt difficulty levels in musical learning tasks

(Yuksel et al., 2016); and challenge and increase

exertion in players during exercising with

videogames (exergames) based on real-time cardiac

responses (Stach et al., 2009). The use of BL

adaptations in videogames has shown that stress,

boredom, enjoyment, anxiety, engagement,

concentration, and alertness can be effectively used

to improve the overall game user experience

(Bontchev, 2016). Despite the increasing popularity

of BLs among game designers and game user

researchers (Pope et al., 2014), its implementation

still faces several limitations regarding the

integration of physiological sensors, the processing

of signals, and the communication between

physiological systems and videogames (Novak,

2014).

In this paper, we present the development of the

Biocybernetic Loop Engine (BL Engine), a flexible

and integrated software tool (from sensing to

decision making) to create BLs. Our solution is

technology agnostic and can be integrated into any

existing software platform. The BL Engine builds on

Muñoz, J., Gouveia, E., Cameirão, M. and Badia, S.

The Biocybernetic Loop Engine: An Integrated Tool for Creating Physiologically Adaptive Videogames.

DOI: 10.5220/0006429800450054

In Proceedings of the 4th International Conference on Physiological Computing Systems (PhyCS 2017), pages 45-54

ISBN: 978-989-758-268-4

top of a solid BL theoretical construct (Serbedzija

and Fairclough, 2009) and proposes a more practical

and applied adaptation technique. We start by

describing the available software platforms that can

be used to construct BLs highlighting some of their

characteristics, advantages and limitations; then we

introduce our BL Engine framework, including

software design and implementation processes.

Finally, a proof-of-concept experiment is presented

using heart rate (HR) based adaptation in a

cardiorespiratory fitness exergame.

2 RELATED WORK

The development of BLs has been advanced by

academia mainly for research purposes (Pope et al.,

2014). Several examples have demonstrated the

efficacy of physiological adaptation to improve

system automation (Prinzel III et al., 2003), player

engagement in gaming experiences (Ewing et al.,

2016) and exertion levels in fitness interventions

(Ketcheson et al., 2015). Although BLs enable the

creation of genuine intelligent systems that use

implicit task-context and user-intention information

(Jacucci et al., 2015), the creation of such systems is

inherently difficult. The fundamental architecture of

BLs requires a systematic integration of human-

body signals, data conditioning for artefact removal

and noise reduction, a feature extraction stage, and a

psychophysiological inference process to finally

translate data to action (Fairclough and Gilleade,

2012). As a result, this is an arduous process and

most of the time the construction of BLs is custom-

built for single-task systems, which makes it

difficult to replicate or generalize to other

applications (Pope et al., 2014). This impedes

researchers and developers to rapidly design,

construct, iterate and validated new prototypes.

Some software tools have emerged in the last

decade to facilitate the creation of BLs and

spreading the use of physiological adaptation for

multiple purposes. Interestingly, all of them use

visual language scripting techniques to simplify the

construction process. One well-known example is

the OpenViBE software platform (Renard et al.,

2010), an open-sourced tool created to support

brain-computer interface (BCI) experiments. Using a

modular, flexible and simplistic architecture,

OpenViBE has been successfully used in closed-

loop systems for assistive technology such as

spellers, as well as for BCI videogames and virtual

reality simulations (Clerc et al., 2016)

(Vourvopoulos et al., 2015). Although OpenViBE

has been mainly used for BCI applications, studies

using ECG data for tangible interfaces (Gervais et

al., 2016) showed the potential of the software

besides neurophysiological signals. The FlyLoop

framework (Peck et al., 2015) is a small and

lightweight approach in Java that enables

programmers to rapidly develop and experiment

with physiologically intelligent systems. The system

is presented as a tool to improve decision-making in

workload detection via wearable biosensors.

Consisting of a set of four modules (data sources,

filters, learners, and outputs), the framework is

designed to provide reproducibility and accessibility

to non-programmer users. Finally, the Neuromore

platform was initially designed as a flexible tool to

create novel biofeedback visualizations (Jillich,

2014) (Kosch et al., 2016). Nowadays the tool is

presented as a development platform for interactive

applications which can combine real time

physiological data and machine intelligence to create

BLs. Focused in the use of wearable and low-cost

BCI systems, Neuromore combines multiple

technologies to connect commercial-grade

physiological sensors with visual scripting. It can

process data and classify it in terms of states of mind

such as focus, relaxation, flow, creativity or

concentration. Unfortunately, the software is still in

early stages and the integration with game engines is

still unclear.

Some features are still lacking in current

software tools to create physiologically adaptive

videogames, specifically: a) versatility to support

multiple body signals; b) integration with game

engines; and c) simplicity to create adaptation rules.

To tackle the limitations of the existing technologies

we have developed the Biocybernetic Loop Engine.

3 THE BIOCYBERNETIC LOOP

ENGINE FRAMEWORK

3.1 Software Design and Development

3.1.1 Design Requirements

The BL engine software was designed to be used by

people both with and without specific training in

physiological computing and/or programming skills.

We identified a list of implementation requirements

that guided the design of the BL Engine in the signal

acquisition, signal processing and feature extraction,

and adaptation domains, as well as its integration

with other software systems.

PhyCS 2017 - 4th International Conference on Physiological Computing Systems

Figure 1: The BL Engine system architecture, which covers the signal acquisition stage via supporting multiple sensing

devices and allowing a basic processing of the signals to extract the HR data. It includes a console to create adaptive rules

and facilitates the communication with videogames, natively supporting those developed in Unity3D.

Signal Acquisition, Signal Processing and Feature

Extraction: one of the biggest limitations when using

physiological signals in interactive projects is the

connectivity with multiple devices. The lack of

standardization of components, different communi-

cation protocols and measurements offer a highly

variable scenario (Novak, 2014), thus the BL engine

should facilitate and streamline the signal acquisition

process. Further, the real time signal processing of the

acquired signals is an engineering challenge (Jacucci

et al., 2015), and thus, the inclusion of common filters

to process signals is imperative. Finally, although the

features commonly extracted from physiological

signals to carry out psychophysiological inferences

are relatively well-defined (Cowley et al., 2016), their

use for biocybernetic adaptation is still not well

understood. Thus, the extraction of meaningful

physiological parameters from sensor signals is

necessary as they are the main input of the BLs.

Adaptation: the second set of requirements relates to

the design of the adaptive rules, which contain the

intelligence of the BL system. Essentially, these

rules encompass the decision-making process

underlying physiological adaptation. Although

simple Boolean rules based on if/then rules have

been successfully used in past investigations (Karran

et al., 2015), more advanced techniques based on

proportional-integral-derivative control (Parnandi et

al., 2013) and machine learning approaches

(Verhulst et al., 2015) have also shown encouraging

results. Despite those advances, the implementation

and iteration of adaptive rules in BLs require

extensive reprogramming processes in order to

create playable prototypes (Pope et al., 2014).

Consequently, our BL engine should embrace an

agile methodology that facilitates the generation of

adaptive rules and enables a fast iteration on them.

Integration: finally, a full integration with third

party software systems, such as videogames, is

required. Even though excellent game engines are

freely available (e.g. Unity3D, Unreal Engine), the

integration of physiological computing technologies

in those systems is not a simple task due to the lack

of standardized and functional signal processing

toolboxes (Bontchev, 2016). Only few examples

enable the integration of physiological sensors with

the Unity3D game engine such as the PhysSigTK

(Rank and Lu, 2015), RehabNet CP (Vourvopoulos

et al., 2013) and PhysioVR framework (Muñoz et

al., 2016). However, BLs require not only a simple

integration of sensors but also a bi-directional

communication between the extracted physiological

parameters and the videogame variables in real time.

3.1.2 Design Process

The BL Engine aims to be an extensive tool for the

creation of BLs in multiple dimensions such as

cardiac, muscular, emotional or motor domains. At

this stage we developed the cardiac module and tested

its functionality implementing a BL in gaming

applications. We used multiple techniques from

software engineering for the development - process

workflow understanding, activities and system

dynamics visualization relying on flow and UML

diagrams of the system, and low fidelity prototyping

through digital interactive wireframes-prior to its

implementation.

The Biocybernetic Loop Engine: An Integrated Tool for Creating Physiologically Adaptive Videogames

Figure 2: Screenshot of the Signal Acquisition Panel in the BL Engine. The image shows an ECG signal from the

CardioBan chest strap (PLUX, Lisbon, Portugal) with the computed HR. Options for band-pass filtering and outliers’

detection are activated to improve the HR computation.

3.1.3 Implementation

The BL Engine is a software tool implemented in

Unity 3D (Unity Technologies, San Francisco, USA)

and it is composed by 3 main modules: a) the signal

acquisition panel, b) the biocybernetic console, and

c) the game connector. Using the BL Engine, users

are able to easily design physiological adaptations of

their videogames following the complete processing

pipeline from physiological data collection, analysis

to the final translation in videogames (see figure 1).

Signal Acquisition Panel: the BL Engine supports

the acquisition of a basic range of wearable devices

(figure 2) including the Biosignal Plux (PLUX,

Lisbon, Portugal), a professional biosignal

acquisition kit with 8-12 bit resolution and 1000Hz

sampling rate that measures blood volume pressure

through photopletysmography (PPG),

electrocardiography (ECG), electromyography

(EMG), electrodermal activity (EDA), and

respiration; a chest strap sensor called CardioBan

(PLUX, Lisbon, Portugal) with integrated ECG,

respiration and acceleration sensors; Bitalino

(PLUX, Lisbon, Portugal), a low cost DIY biosignal

board with ECG, EMG, light intensity, and

acceleration sensing; HR data streamed from the

RealSense Intel camera (Intel, California, USA); and

HR data through Android Wear devices such as

smartwatches and wristbands through the PhysioVR

App (Muñoz et al., 2016). UDP is used for the

communication of the Bioplux, CardioBan and the

android wearables through external applications

streaming it in the RehabNet protocol

(Vourvopoulos et al., 2013), while the Bitalino

integration is natively supported by the BL Engine

through a serial port interface over a bluetooth

connection.

The signal acquisition panel also includes a signal

visualization to facilitate the real time data analysis

and the feature extraction from data. Two cardiac-

related signals can be processed at this stage: PPG

and ECG. The acquisition panel includes algorithms

for the HR computation based on an adaptive peak-

detection technique. Both the peak width and the

peak height can be manually adjusted, and

adjustable band-pass filters can be used to improve

the accuracy of the HR computation. Finally, after

the HR computation, an outlier detection algorithm

is used based on the following statistical descriptors:

 = ( − 



) > ( ∗ ())

(1)

where X are the HR measurements in a temporal

window, 



is the mean value of X, and () the

standard deviation of X. zFactor is a constant with a

default value of 2, meaning that every HR data point

that differs by more than 2 standard deviations from

the previous data point will be considered an outlier.

The filtered HR data is then sent to the

Biocybernetic Console for further processing.

PhyCS 2017 - 4th International Conference on Physiological Computing Systems

Figure 3: A screenshot of the Biocybernetic Console in the BL Engine representing an adaptation rule in which the game

Variable A is increased by a value of one (1) once the HR is between 80 and 100 BPMs.

Biocybernetic Console: this console is designed to

facilitate the construction of the adaptive rules that

use the output computed by the signal acquisition

panel to influence the videogame functioning (figure

3). This is achieved through a visual scripting

module, which comprises the use of pre-

programmed boxes that can be graphically

connected to create adaptive rules through functional

physiological pipelines (FPP). The blocks can be

dragged-and-dropped from the right-side canvas to

the left workspace, and inputs and outputs of the

boxes can be connected drawing connecting lines.

Additionally, the workspace size can be modified

allowing the creation of multiple FPPs that can run

in parallel using inputs from different physiological

sensors. The blocks fall into three different

categories:

• Receivers: blocks that receive and/or simulate

data. Here, we can even receive data directly

from the signal acquisition panel or data coming

from any external application supporting socket

programming through the UDP Reh@Net

protocol.

• Operators: blocks that make comparisons,

mathematical and logical operations, and

variables’ assignments. Blocks for adding

constants and visualizing results are also

implemented.

• Game outputs: blocks for modifying game

variables in real time. Game variables are

exposed to the BL Engine using the Unity3D

Game Connector module or through UDP for

third-party applications.

Finally, to expand the possibilities of creating

adaptation rules, the biocybernetic console includes

a JavaScript interpreter that allows the generation of

more complex adaptation rules directly coding them

and executing them on-the-fly. After the rule

creation process, users can test its behaviour in real-

time and iterate with multiple adaptive rules during

run-time. Data from both the BL Engine and the

videogame can be synchronously recorded for post-

processing using a CSV data writer script.

Game Connector: to enable the connectivity

between the BL Engine and the videogames, we

provide the game connector module, which is

wrapped into a prefabricated package (prefab) that

can be integrated in any videogame developed in

Unity 3D. The Unity prefab package contains the

scripts needed for bidirectional communication with

the BL Engine. The connector receives the

physiological data via UDP communication, makes

specific videogame variables available to the

Biocybernetic Console for the creation of the

adaptation rules, and updates them in real time

accordingly. Any third party application supporting

socket programming (such as Unreal Engine and

others) can also receive data from the biocybernetic

console via a UDP parsing module.

The Biocybernetic Loop Engine: An Integrated Tool for Creating Physiologically Adaptive Videogames

4 PRELIMINARY ASSESSMENT

With the objective of assessing the feasibility of

using the BL Engine technology to create

physiologically adaptive videogames, we presented

a case study using HR data in an Exergaming

experience.

4.1 Physiologically Adaptive Exerpong

4.1.1 Exergame Design

Exerpong is an exergame developed in Unity3D and

designed for agility and balance training in active

seniors (Muñoz J.E. et al., 2016). The exergame was

designed as an adaptation of the classic 2D Pong in

which the goal is to hit a ball using a virtual paddle.

We used the BL Engine to make adaptations based

on the HR of users during Exerpong gameplay, with

the goal of driving their HR to a target zone. The

target HR zone is expressed in terms of the heart rate

reserve (HRR) which is the difference between the

maximum HR and the HR during the resting state. In

this zone, the health benefits of a cardiorespiratory

training session can be maximized via stressing the

cardiac muscle without over-exercised it (Heyward

and Gibson, 2014). For older adults, the ACSM

recommends exercise at 40% to 70% of the HRR for

moderate intensities (Rahl, 2010) during sessions of

20 to 30 minutes.

Game parameters are adjusted following the dual

flow model for exergaming (Sinclair et al., 2009), in

our case to adapt for Gameplay and

CardioRespiratory Fitness. Gameplay adaptation

changes games parameters to improve game

attractiveness and balance the challenge in the

following way:

• The paddle size decreases once the player hits

the ball and increases once he/she misses it.

• The ball velocity decreases if the player misses

three consecutive balls.

The CardioRespiratory Fitness component of the

Exerpong is adapted using the real time HR data

according to the following rule:

• The ball velocity increases if the 30-seconds HR

average is under 50% of the HRR, and

decreases otherwise.

4.1.2 Experimental Setup

A white 2.5m x 3.0m PVC surface was used to

project Exerpong on the floor. The KinectV2 sensor

(Microsoft, Washington, USA) was used for

tracking the user’s waist position and mapping it to

the paddle position. A smartwatch Motorola 360

Sport was used to capture HR data at 1 Hz sampling

frequency as input for the physiological adaptation.

ECG signals were recorded using the Biosignal Plux

at 1000 Hz through a triode dry electrode placed on

the V

pre-cordial derivation. An extra-elastic band

was used to reinforce the contact and stability

between the electrodes and the skin during the

exercise. The figure 4 shows the experimental setup.

Figure 4: Diagram illustrating the experiment setup for the

adaptive Exerpong consisting of a KinectV2 sensor, a

projected environment and the physiological sensors.

The HR computed via the post-processing of

ECG signals was used as a ground-truth and

compared with the smartwatch data.

4.2 Case Study

A community-dwelling 62-years-old female

participated in the study. The participant was

recruited at a local senior sports facility. The

Montreal Cognitive Assessment (MoCA) (Freitas et

al., 2011) was used for cognitive screening. The

participant scored 29/30, indicating normal cognitive

function to understand the instructions and

participate in the experiment. The level of physical

activity was assessed using the short version of the

International Physical Activity Questionnaire

(IPAQ) (Booth et al., 2003), through which the user

was categorized for moderate intensity of physical

activity. The body mass index was 24 kg/m

indicating normal weight. The HR during a 5

minutes resting period was calculated as 71 BPMs,

while the HR maximum was estimated to be 164

BPMs following Tanaka’s formula (Tanaka et al.,

2001). Then, the target HR value was established as

117 BPMs (50% of HRR).

4.3 Protocol

After arrival, the participant received the informa-

PhyCS 2017 - 4th International Conference on Physiological Computing Systems

tion about the study, signed an informed consent,

provided the demographic information, and

underwent IPAQ and MoCA assessments.

Subsequently, the participant was asked to remain 5

minutes seated for collection of HR data during

resting. ECG signals and the HR from the

smartwatch were collected synchronously. A short

stretching routine was used to facilitate the muscle

and tendons exertion of the lower limbs. The game

mechanics of the Exerpong were explained before

starting. The value for ball velocity started from the

minimum and changed every 30 seconds following

the BL Engine adaptations described before. Initial

parameters of the user such as age, HR during

resting and target HR percentage were configured

for the adaptation. The interaction with the adaptive

Exerpong lasted 20 minutes.

4.4 Results

In order to validate the use of the smartwatch as

real-time input to the BL Engine, the post-processed

HR data from the ECG signals was used for

comparison with the data from the smartwatch. A

low root-mean-square-error of 5 BPMs was

computed, which is in accordance with previously

reported values (Mike Prospero, 2016).

Figure 5 shows the resulting Gameplay and

CardioRespiratory Fitness adaptations during

exercising with the physiologically adaptive

Exerpong. Figure 5A shows the HR measurements

of the user during the complete session, and the red

line indicates the target HR value (117 BPMs). We

can observe from the data that after approximately

six minutes of training the user reached the expected

value. Figure 5B shows that this happens as a result

of a constant increase in ball velocity. Through

playing ExerPong, the user achieved an average HR

of 116 BPMs, very close to the intended target HR.

This value is considerably higher than the registered

average HR in a conventional training session in the

same senior gym, which is 93 BPMs.

Figure 5B shows the dynamics of the

CardioRespiratory Fitness adaptations by the BL

Engine to modify the ball velocity. The result is a

classic bang-bang controller switching between +1

and -1 values depending on the HR value relative to

the target. It can be observed that despite the binary

decisions, the BL Engine successfully engaged the

user and modulated her HR oscillating around the

target value, crossing it multiple times (minutes 6,

10, 12, 14, 18) during the session as reaction to the

game adaptations. Finally, Figure 5C shows the

behaviour of the two game variables that were

Figure 5: Adaptation during the session with Exerpong. A:

HR responses (blue line) and target heart rate (red line); B:

adaptation values (+1 or -1) sent from the BL Engine to

increase or decrease the ball velocity; C: behavior of both

paddle size and ball velocity game parameters.

modified by both Gameplay and CardioRespiratory

Fitness adaptations. It can be observed that the ball

speed (blue line) acted as the main driver for the

changes in HR values. A cross-correlation analysis

of those two variables showed a very high similarity

(0.82). Moreover, the paddle size (red line) was

being reduced by the Gameplay adaptations, making

the task more challenging and encouraging

movement performance, thus facilitating the

modulation of HR responses.

5 DISCUSSION & CONCLUSION

This paper presented the design, implementation and

a proof-of-concept of the Biocybernetic Loop

Engine tool, which is freely available at:

http://neurorehabilitation.m-iti.org/tools/blengine.

The here presented proof-of-concept experiment

evaluated the feasibility of including BLs to adapt

exergaming experiences for the maximization of its

effectiveness for cardiorespiratory training in

seniors. This is maybe one of the more complex

scenarios for real-time HR-based adaptation, since

monitoring cardiac responses during exercise is

particularly challenging due to movement artefacts.

The use of tools such as the BL Engine will

facilitate a better understanding of the role of BLs in

The Biocybernetic Loop Engine: An Integrated Tool for Creating Physiologically Adaptive Videogames

gaming technologies, a more streamlined

connectivity with physiological sensors, a fast

iteration of adaptation techniques, and an easy

integration of physiological intelligence in

videogames. The BL Engine addresses such

challenges through relatively low-cost and wearable

physiological sensors such as smartwatches, utilizes

a fully functional and modular user interface,

integrates a visual scripting module, which

facilitates programming the adaptation rules, and

provides tools for a simplistic integration of any

videogame developed in Unity3D. Furthermore, the

system provides a integrated and comprehensible

architecture which might facilitate the incorporation

of multiple physiological features streamed from

several sensors and captured in the biocybernetic

console, hence permitting the conception of

multimodal BLs (D’mello and Kory, 2015).

Novel physiologically modulated videogames

might overcome the existing limitations and become

part of our daily activities through systems such as

exergames for exercise prescription or interactive

applications for stress management. This will bring

uncountable benefits in augmenting human

computer interactions. Through the integration of

physiological adaptation, more affective and

personalized videogames can be developed enabling

a fluent communication between the physiological

parameters and the videogame variables.

6 FUTURE WORK

Currently, we are planning a cross-sectional study

with the adaptive Exerpong in a group of senior

users for evaluating the appropriateness of the

adaptation for boosting effectiveness in exergaming-

based interventions for exercise promotion. In

addition, a longitudinal intervention will be

conducted, to compare the effectiveness of such

approach in comparison to traditional physical

exercise activities.

An interesting future application of the BL

Engine could be the creation of adaptive rules based

on heart rate variability (HRV) analysis using

specific measurements which have been associated

to workload and stress. For instance, novel serious

videogames for stress management can react

dynamically to HRV parameters such as the SDNN

(standard deviation of normal R-R intervals). This

will provide a very compelling scenario for train the

awareness of one’s internal physiological states (also

called interoceptive awareness) (Schulz and Vögele,

2015), which might be one of the cornerstones in the

wellbeing upsurge via physiological computing

technologies (Critchley et al., 2004). Finally, we

believe that the BLs have high potential for being

integrated with virtual reality applications, opening

up a new communication pathway for adaptive

contents creation (Siriborvornratanakul, 2016).

Although the BL Engine only contains the

cardiorespiratory module at this stage, the simplified

data collection and analysis and translation model

can be transversally used for physiological

adaptations besides cardiac-related signals.

7 CONTRIBUTIONS

JEM and SBB defined and designed the BL Engine.

JEM implemented the software. All authors defined

the experimental protocol. JEM collected and

analyzed the data. All authors interpreted the results.

All authors revised and approved the current version

of the manuscript.

ACKNOWLEDGEMENTS

The authors would like to thank Teresa Paulino for

developing the Exerpong, and for contributing to the

development of the signal acquisition panel, the

game connector, and the final integration of the

system; and Luis Quintero for contributing to the

development of the visual scripting module for the

biocybernetic console. This work was supported by

the Portuguese Foundation for Science and

Technology through the Augmented Human

Assistance project (CMUP-ERI/HCI/0046/2013),

Projeto Estratégico UID/EEA/50009/2013, and

ARDITI (Agência Regional para o Desenvolvimento

da Investigação, Tecnologia e Inovação).

REFERENCES

Bontchev, B., 2016. Adaptation in Affective Video

Games: A Literature Review. Cybern. Inf. Technol. 16,

3–34.

Booth, M. L., Ainsworth, B. E., Pratt, M., Ekelund, U.,

Yngve, A., Sallis, J. F., Oja, P., 2003. International

physical activity questionnaire: 12-country reliability

and validity. Med Sci Sports Exerc 195, 3508–1381.

Clerc, M., Bougrain, L., Lotte, F., 2016. Brain-Computer

Interfaces 2: Technology and Applications. John

Wiley & Sons.

Cowley, B., Filetti, M., Lukander, K., Torniainen, J.,

Henelius, A., Ahonen, L., Barral, O., Kosunen, I.,

PhyCS 2017 - 4th International Conference on Physiological Computing Systems

Valtonen, T., Huotilainen, M., others, 2016. The

Psychophysiology Primer: A Guide to Methods and a

Broad Review with a Focus on Human–Computer

Interaction. Found. Trends® Human–Computer

Interact. 9, 151–308.

Critchley, H. D., Wiens, S., Rotshtein, P., Öhman, A.,

Dolan, R. J., 2004. Neural systems supporting

interoceptive awareness. Nat. Neurosci. 7, 189–195.

D’Mello, S. K., Kory, J., 2015. A review and meta-

analysis of multimodal affect detection systems. ACM

Comput. Surv. CSUR 47, 43.

Ewing, K. C., Fairclough, S. H., Gilleade, K., 2016.

Evaluation of an Adaptive Game that Uses EEG

Measures Validated during the Design Process as

Inputs to a Biocybernetic Loop. Front. Hum. Neurosci.

10.

Fairclough, S., Gilleade, K., 2012. Construction of the

Biocybernetic Loop: A Case Study, in: Proceedings of

the 14th ACM International Conference on

Multimodal Interaction, ICMI ’12. ACM, New York,

NY, USA, pp. 571–578. doi: 10.1145/2388676.

2388797.

Fairclough, S. H., 2009. Fundamentals of physiological

computing. Interact. Comput. 21, 133–145.

Fairclough, S. H., Gilleade, K., 2014. Advances in

physiological computing. Springer.

Freitas, S., Simões, M.R., Alves, L., Santana, I., 2011.

Montreal Cognitive Assessment (MoCA): normative

study for the Portuguese population. J. Clin. Exp.

Neuropsychol. 33, 989–996.

Gervais, R., Frey, J., Gay, A., Lotte, F., Hachet, M., 2016.

Tobe: Tangible out-of-body experience, in:

Proceedings of the TEI’16: Tenth International

Conference on Tangible, Embedded, and Embodied

Interaction. ACM, pp. 227–235.

Gilleade, K., Dix, A., Allanson, J., 2005. Affective

Videogames and Modes of Affective Gaming: Assist

Me, Challenge Me, Emote Me. Presented at the

DiGRA 2005: Changing Views – Worlds in Play.

Heyward, V. H., Gibson, A., 2014. Advanced Fitness

Assessment and Exercise Prescription 7th Edition.

Human Kinetics.

Jacucci, G., Fairclough, S., Solovey, E.T., 2015.

Physiological Computing. Computer 48, 12–16.

doi:10.1109/MC.2015.291.

Jillich, B., 2014. Acquisition, analysis and visualization of

data from physiological sensors for biofeedback

applications.

Karran, A. J., Fairclough, S. H., Gilleade, K., 2015. A

framework for psychophysiological classification

within a cultural heritage context using interest. ACM

Trans. Comput.-Hum. Interact. TOCHI 21, 34.

Ketcheson, M., Ye, Z., Graham, T.C.N., 2015. Designing

for Exertion: How Heart-Rate Power-ups Increase

Physical Activity in Exergames, in: Proceedings of the

2015 Annual Symposium on Computer-Human

Interaction in Play, CHI PLAY ’15. ACM, New York,

NY, USA, pp. 79–89. doi:10.1145/2793107.2793122.

Kosch, T., Hassib, M., Schmidt, A., 2016. The Brain

Matters: A 3D Real-Time Visualization to Examine

Brain Source Activation Leveraging Neurofeedback,

in: Proceedings of the 2016 CHI Conference Extended

Abstracts on Human Factors in Computing Systems.

ACM, pp. 1570–1576.

Loeb, R. G., Cannesson, M., 2017. Closed-Loop

Anesthesia: Ready for Prime Time? LWW.

Mike Prospero, 2016. Who Has The Most Accurate Heart

Rate Monitor? Tomsguide.

Mishra, J., Gazzaley, A., 2014. Closed-loop rehabilitation

of age-related cognitive disorders, in: Seminars in

Neurology. Thieme Medical Publishers, pp. 584–590.

Muñoz J. E., Bermudez S., Rubio E., Cameirao M., 2016.

Modulation of Physiological Responses and Activity

Levels During Exergame Experiences, in: 2016 18th

International Conference on Virtual Worlds and

Games for Serious Applications. IEEE, p. In press.

Muñoz, J. E., Paulino, T., Vasanth, H., Baras, K., 2016.

PhysioVR: A novel mobile virtual reality framework

for physiological computing, in: E-Health Networking,

Applications and Services (Healthcom), 2016 IEEE

18th International Conference on. IEEE, pp. 1–6.

Novak, D., 2014. Engineering Issues in Physiological

Computing, in: Advances in Physiological Computing.

Springer, pp. 17–38.

Novikov, D. A., 2016. Cybernetics in the 20th Century, in:

Cybernetics. Springer, pp. 1–19.

Parnandi, A., Son, Y., Gutierrez-Osuna, R., 2013. A

Control-Theoretic Approach to Adaptive Physiolo-

gical Games, in: 2013 Humaine Association

Conference on Affective Computing and Intelligent

Interaction. Presented at the 2013 Humaine

Association Conference on Affective Computing and

Intelligent Interaction, pp. 7–12. doi: 10.1109/ACII.

2013.8.

Peck, E. M., Easse, E., Marshall, N., Stratton, W., Perrone,

L. F., 2015. FlyLoop: a micro framework for rapid

development of physiological computing systems, in:

Proceedings of the 7th ACM SIGCHI Symposium on

Engineering Interactive Computing Systems. ACM, pp.

152–157.

Pope, A. T., Bogart, E. H., Bartolome, D. S., 1995.

Biocybernetic system evaluates indices of operator

engagement in automated task. Biol. Psychol. 40, 187–

195.

Pope, A. T., Stephens, C. L., Gilleade, K., 2014.

Biocybernetic Adaptation as Biofeedback Training

Method, in: Fairclough, S. H., Gilleade, K. (Eds.),

Advances in Physiological Computing, Human–

Computer Interaction Series. Springer London, pp.

91–115.

Prinzel III, L. J., Parasuraman, R., Freeman, F. G., Scerbo,

M. W., Mikulka, P.J., Pope, A.T., 2003. Three

experiments examining the use of electroencepha-

logram, event-related potentials, and heart-rate

variability for real-time human-centered adaptive

automation design.

Rahl, R. L., 2010. Physical activity and health guidelines.

Recomm. Var. Ages Fit. Levels Cond. From 57.

Rank, S., Lu, C., 2015. PhysSigTK: Enabling engagement

experiments with physiological signals for game

The Biocybernetic Loop Engine: An Integrated Tool for Creating Physiologically Adaptive Videogames

design, in: Affective Computing and Intelligent

Interaction (ACII), 2015 International Conference on.

IEEE, pp. 968–969.

Renard, Y., Lotte, F., Gibert, G., Congedo, M., Maby, E.,

Delannoy, V., Bertrand, O., Lécuyer, A., 2010.

OpenViBE: an open-source software platform to

design, test, and use brain-computer interfaces in real

and virtual environments. Presence Teleoperators

Virtual Environ. 19, 35–53.

Schulz, A., Vögele, C., 2015. Interoception and stress.

Front. Psychol. 6, 993.

Serbedzija, N. B., Fairclough, S.H., 2009. Biocybernetic

loop: From awareness to evolution, in: Evolutionary

Computation, 2009. CEC’09. IEEE Congress on.

IEEE, pp. 2063–2069.

Sinclair, J., Hingston, P., Masek, M., 2009. Exergame

development using the dual flow model, in:

Proceedings of the Sixth Australasian Conference on

Interactive Entertainment. ACM, p. 11.

Siriborvornratanakul, T., 2016. A Study of Virtual Reality

Headsets and Physiological Extension Possibilities, in:

International Conference on Computational Science

and Its Applications. Springer, pp. 497–508.

Stach, T., Graham, T. C., Yim, J., Rhodes, R.E., 2009.

Heart rate control of exercise video games, in:

Proceedings of Graphics Interface 2009. Canadian

Information Processing Society, pp. 125–132.

Tanaka, H., Monahan, K. D., Seals, D.R., 2001. Age-

predicted maximal heart rate revisited. J. Am. Coll.

Cardiol. 37, 153–156.

Verhulst, A., Yamaguchi, T., Richard, P., 2015.

Physiological-based Dynamic Difficulty Adaptation in

a Theragame for Children with Cerebral Palsy., in:

PhyCS. pp. 164–171.

Vourvopoulos, A., Cardona, J.E.M., Bermudez i Badia, S.,

2015. Optimizing motor imagery neurofeedback

through the use of multimodal immersive virtual

reality and motor priming, in: Virtual Rehabilitation

Proceedings (ICVR), 2015 International Conference

on. IEEE, pp. 228–234.

Vourvopoulos, A., Faria, A. L., Cameirão, M.S.,

Bermudez i Badia, S., 2013. RehabNet: A distributed

architecture for motor and cognitive neuro-

rehabilitation, in: 2013 IEEE 15th International

Conference on E-Health Networking, Applications

Services (Healthcom). pp. 454–459.

Yuksel, B. F., Oleson, K.B., Harrison, L., Peck, E.M.,

Afergan, D., Chang, R., Jacob, R.J., 2016. Learn piano

with BACh: An adaptive learning interface that adjusts

task difficulty based on brain state, in: Proceedings of

the 2016 CHI Conference on Human Factors in

Computing Systems. ACM, pp. 5372–5384.

PhyCS 2017 - 4th International Conference on Physiological Computing Systems