FIELD OF THE DISCLOSURE

[0001] The present disclosure relates to discordant sensory stimulus in virtual realitybased exercise and in particular a virtual reality exercise system and methods and a controller for control thereof.

BACKGROUND OF THE DISCLOSURE

[0002] Virtual reality-based exercise has become increasingly common in recent years due to improvements in its sophistication and realism, the wider availability and decrease in cost of the required apparatus, and also the growing trend for home-based exercise.

[0003] Virtual reality-based exercise allows a user to replicate in a virtual world a wide range of conditions, environments and locations that they may not normally have access to, and also to control specific exercise parameters so that they may individually tailor the virtual reality-based exercise to their specific needs. For example, with respect to virtual realitybased cycling, aspects such as the physical resistance presented to the user through the bike pedals, the gearing available to the user, and the sensory stimuli output to the user may all be controlled.

[0004] However, in addition to replicating real-world conditions, the ability to control the parameters of virtual reality-based exercise enables non-real-world conditions to be simulated, which may allow novel exercise regimes to be created and targeted data collection to be performed.

BRIEF SUMMARY OF THE DISCLOSURE

[0005] According to a first aspect of the present disclosure, there is provided a method for controlling a virtual reality exercise system comprising an exercise apparatus and an output unit for outputting a sensory stimulus to a user, the method comprising: outputting a sensory stimulus through the output unit; controlling a parameter of the exercise apparatus according to a discordant relationship between the parameter of the exercise apparatus and a parameter of the sensory stimulus; detecting a user exercise input through the exercise apparatus and obtaining a measurement of a characteristic of the user; and determining a relationship between the parameter of the sensory stimulus and the characteristic of the user based on the characteristic measurement and the discordant relationship.

[0006] Among other advantages, this method enables the influence of a sensory stimulus on a characteristic of a user during exercise to be determined. In turn, the user characteristic may then be manipulated or controlled via the use of the sensory stimulus, which may for example allow the user to exercise at a higher intensity. [0007] In one example, the parameter of the sensory stimulus corresponds to the parameter of the exercise apparatus. Such a relationship enables the changing of the sensory stimulus to be integrated in a realistic manner with the user’s virtual reality experience. For example, with respect to cycling, the sensory stimulus may be a displayed gradient and the parameter of the exercise apparatus may be a simulated gradient i.e. a resistance presented by the exercise apparatus.

[0008] In one example, the parameter of the exercise apparatus is one of a gradient, a speed, a weight, a head wind, and a water current.

[0009] In one example, the parameter of the sensory stimulus is one of a gradient, speed, weight, head wind, and water current.

[0010] In one example, determining the relationship between the parameter of the sensory stimulus and the characteristic of the user comprises determining a correlation coefficient between the parameter of the sensory stimulus and the characteristic of the user.

[0011] In one example, the outputting of the sensory stimulus includes outputting a sequence of sensory stimuli having different values of the parameter of the sensory stimulus.

[0012] In one example, at least two of the sequence of sensory stimuli have equal values of the parameter of the sensory stimulus and the values of the parameter of the exercise apparatus during the output of each of the at least two sensory stimuli are different.

[0013] In one example, at least two of the sequence of sensory stimuli have different values of the parameter of the sensory stimulus and the values of the parameter of the exercise apparatus during the output of each of the at least two sensory stimuli are equal.

[0014] In one example, outputting the sensory stimulus includes outputting visual virtual reality content.

[0015] In one example, the method further comprises updating the sensory stimulus based on the received exercise input.

[0016] In one example, the exercise apparatus is an ergometer.

[0017] In one example, the exercise apparatus is one of a stationary bicycle, a treadmill, a rowing machine, a step machine, an elliptical trainer, a weights machine, and a climbing machine.

[0018] In one example, the characteristic of the user is a physiological characteristic of the user.

[0019] In one example, the characteristic includes at least one of a breathlessness of the user, a breathing rate of the user, a tidal volume of the user, a minute ventilation of the user, a heart rate of the user, a power output by the user through the exercise apparatus, a physical discomfort of the user, a fatigue of the user, and an anxiety of the user.

[0020] In one example, the parameter of the exercise apparatus has an inconsistent discordant relationship with the parameter of the sensory stimulus.

[0021] In one example, the method further comprises updating at least one of the sensory stimulus and the discordant relationship based on the determined relationship.

[0022] According to a second aspect of the present disclosure, there is provided a method for controlling a virtual reality exercise system comprising an exercise apparatus and an output unit for outputting a sensory stimulus to a user, the method comprising: outputting a sensory stimulus through the output unit; controlling a parameter of the exercise apparatus based on a discordant relationship between the parameter of the exercise apparatus and a parameter of the sensory stimulus; detecting a user exercise input through the exercise apparatus; and updating the output sensory stimulus based on the user exercise input.

[0023] Advantageously, the user characteristic may be manipulated or controlled via the use of the sensory stimulus, which may for example allow the user to exercise at a higher intensity.

[0024] In one example, the discordant relationship has been determined based on a relationship between a characteristic of the user and parameter of the sensory stimulus.

[0025] In one example, the relationship between a characteristic of the user and parameter of the sensory stimulus has been determined in accordance with the first aspect.

[0026] According to a third aspect, there is provided a controller for a virtual reality exercise system comprising an exercise apparatus and an output unit for outputting a sensory stimulus to a user, wherein the controller is configured to implement the method of any one of the preceding aspects or examples.

[0027] In accordance with a fourth aspect, there is provided virtual reality exercise system comprising an exercise apparatus, a sensory stimulus output unit, and a controller according to the third aspect.

[0028] In accordance with a fifth aspect, there is provided a computer-readable recording medium having stored thereon instructions which, when executed by a computer, cause the computer to perform the method of any of the first to third aspects and the corresponding examples.

BRIEF DESCRIPTION OF THE DRAWINGS [0029] Embodiments of the present disclosure are further described hereinafter with reference to the accompanying drawings, in which:

[0030] Figure 1 provides a schematic diagram of a virtual reality exercise system in accordance with an example of the present disclosure;

[0031] Figure 2 provides a flow diagram of a method for controlling a virtual reality exercise system in accordance with an example of the present disclosure;

[0032] Figure 3 provides a graph illustrating a discordant relationship between a parameter of a sensory stimulus and a parameter of an exercise apparatus of a virtual reality exercise system in accordance with an example of the present disclosure;

[0033] Figures 4a and 4b provide graphs of a relationship between a physiological characteristic and a parameter of a sensory stimulus parameter for different exercise apparatus parameters in accordance with an example of the present disclosure;

[0034] Figure 5a provides a graph of an estimated correlation coefficient between various physiological characteristics and a parameter of a sensory stimulus in accordance with an example of the present disclosure;

[0035] Figure 5b provides a graph of an estimated correlation coefficient between various physiological characteristics and sensory stimulus;

[0036] Figure 6a provides a graph illustrating a concordant relationship between a parameter of a sensory stimulus and a parameter of an exercise apparatus of a virtual reality exercise system in accordance with an example of the present disclosure;

[0037] Figure 6b provides a graph illustrating a discordant relationship between a parameter of a sensory stimulus and a parameter of an exercise apparatus of a virtual reality exercise system in accordance with an example of the present disclosure;

[0038] Figure 6c provides a graph of an estimated correlation coefficient between breathlessness and physiological characteristics and parameters of a sensory stimulus in accordance with an example of the present disclosure;

[0039] Figure 6d provides a graph of an estimated correlation coefficient between leg discomfort and physiological characteristics and parameters of a sensory stimulus in accordance with an example of the present disclosure; and

[0040] Figure 7 provides a flow diagram of a method for controlling a virtual reality exercise system in accordance with an example of the present disclosure.

DETAILED DESCRIPTION

Virtual Reality-Based Exercise [0041] At a basic level, virtual reality-based exercise comprises a user providing an exercise input through an exercise apparatus and presenting some form of sensory stimulus to the user based on information detected from their exercise input and predefined information on a virtual world in which the exercise is deemed to be taking place in.

[0042] With respect to virtual reality cycling, information detected from an exercise input on a stationary bike may include one or more of speed, cadence and power, whereas for virtual reality running on a treadmill, the detected information may include power, ground contact time and cadence for example. This detected information is then provided to a controller that controls the sensory stimulus such that the sensory stimulus reacts to the user’s input and the user is provided with the feeling of interacting with the virtual world. For example, based on a detected power and information on (i.e. parameters of) the virtual world, the controller may determine the user’s speed in the virtual world and update the sensory stimulus accordingly. More advanced forms of virtual reality-based exercise may also allow the controller to control parameters of the exercise apparatus such that there is a circular feedback mechanism between the controller and the exercise apparatus, which allows the exercise apparatus and thus the user experience to better reflect the conditions of the virtual world. The parameters of the exercise apparatus will most commonly relate to a physical resistance, speed, weight or other difficulty-related parameter presented to the user but will vary based on the type of exercise and exercise apparatus. For example, with respect to cycling a resistance presented through the pedals may be controlled, and for running an angle and/or speed of the treadmill may be controlled.

[0043] The sensory stimulus presented to the user is most commonly in the form of visual and/or audio content, such as a first person or third person viewpoint of the user or their avatar performing the exercise and associated audio content, such as breathing sounds or environmental cues. However, other forms of sensory stimulus are also possible depending on the nature of the exercise, such as vibration, a position of the exercise apparatus or air flow. For example, with respect to cycling, sensory stimuli such as the angle of the bike and a fan speed may be controlled as well as visual and/or audio content.

[0044] Although a sensory stimulus is predominantly referred to throughout this disclosure, this term is intended to cover plural stimuli, and may also be referred to as characteristics of the virtual world, or characteristics of the virtual content, and is separate to a parameter of the exercise apparatus that affects the difficulty of the exercise. For example, with respect to cycling, the angle of the bike is considered a sensory stimulus since it is does not directly affect the cycling difficulty to a significant degree, whereas for running, the angle of the treadmill is considered a parameter of the exercise apparatus since it directly affects running difficulty. As one may appreciate, an exercise apparatus may have plurality of controllable parameters and a plurality of different sensory stimuli may be presented to the user in order to present the virtual world to the user.

[0045] Figure 1 provides a schematic diagram of an example virtual reality exercise system 100 in accordance with the present disclosure, where the virtual reality exercise system 100 comprises an exercise apparatus 102, a controller 104, and a sensory stimulus output unit 106.

[0046] The exercise apparatus 102 may be any exercise apparatus that can detect and provide feedback on a user exercise input (i.e. an ergometer) and/or receive control information from a controller for adjusting a parameter of the exercise apparatus. Such exercise apparatus may for example include, but are not limited to, a stationary bike (e.g. an exercise bike or turbo trainer-mounted bike), a treadmill, a rowing machine, a weights machine, a step machine, an elliptical trainer, an arm-based exercise machine, and a climbing machine. Parameters of the exercise apparatus, such as a resistance, speed, weight may be manually or automatically controlled. Exercise apparatuses that are capable of being controlled via automated control information received form an external controller, such as a computer for instance, are often referred to as a smart exercise apparatus.

[0047] The exercise apparatus 102 may include one or more sensors for detecting and providing information of an exercise input (i.e. exercise information) of a user to the controller, where the nature of the sensors will be dependent on the type and sophistication of the exercise apparatus. For example, when the exercise apparatus is a bike, the sensors may include one or more of a wheel speed sensor, a cadence sensor, and a power sensor, whereas for a weights machine a force sensor and position sensors may be more appropriate, and for a treadmill, sensors suitable for detecting power, ground contact time and cadence for example may be provided.

[0048] The controller 104 or a computer program running thereon receives exercise information on the exercise being performed through the exercise apparatus and determines the sensory stimulus to output via the sensory stimuli output unit based on the exercise information and parameters of the virtual world the exercise is taking place in. For example, with respect to cycling and a video-based sensory stimulus, based on exercise information such as a user power and a virtual-world parameter such as a gradient of a hill in the virtual word, the controller may calculate a virtual speed at which the user is cycling and update the video output through the sensory stimulus output unit such that the user is shown to be cycling up the virtual hill at the appropriate speed.

[0049] The controller may also provide control information to the exercise apparatus in order to control one or more parameters of the exercise apparatus, such as a resistance, speed, or weight associated with the exercise apparatus for example. The controller may also perform analysis based on one or more of received exercise information, parameters of the exercise apparatus, and parameters of the virtual world. More details on such analysis are described below.

[0050] The controller may take the form of a multi-purpose computer, such as a desktop computer, a portable computer, a tablet computer, or smartphone, but may also take the form of a dedicated controller that is configured specifically for the role of the controller. When the controller is a multi-purpose computer, a computer program may be executed to control the virtual reality exercise system. For example, with respect to cycling and running Zwift™ is one of a number of applications that may be used to control a virtual reality exercise system.

[0051] The sensory stimulus output unit 106 is arranged to output one or more forms of sensory stimuli to the user whilst they are performing exercise through the exercise apparatus, where the output unit 106 receives the sensory stimulus content from the controller. The sensory stimulus content may take any appropriate form and thus the form of the output unit 106 may vary depending on the sophistication of the virtual reality exercise system and the type of exercise apparatus. Although the most common form of output unit is envisaged to be a visual display unit such as a head-mounted display, a television, or a screen of a computer, it may also take the form of or include speakers for outputting audio, a fan for simulating air flow, a mechanical unit for outputting changes in apparatus position or vibrations for example or any combination of these. Consequently, in the context of this disclosure, virtual reality is considered to be an approach where at least one sensory stimulus corresponding to a virtual world is presented to the user. For example, although a head-mounted display may provide a more immersive experience, the use of a simple display unit such as a television or monitor is also suitable for implementation of the virtual reality experience. Although illustrated as a single entity, there may be multiple output units, for example, with respect to cycling, there may be a display screen, a speaker, a fan, and a mechanical unit to adjust an angle of the bike to simulate the position of the bicycle when it is on a gradient. Each of the output units may be controlled based on sensory stimulus control information received form the controller, where the control information is determined based on exercise information received from the exercise apparatus and/or the parameters of the virtual world.

[0052] Although the exercise apparatus 102, the controller 104 and the output unit 106 are shown as separate entities in Figure 1 , they may also take be combined in any appropriate combination. For example, the controller 104 and the output unit 106 may be combined as a single entity, such as a laptop computer or tablet. In another example, the output unit 106 may be combined with the exercise apparatus 102.

[0053] In some examples the virtual reality exercise system 100 may also have one or more sensors 108 arranged to monitor a physiological characteristic of the user, such as heart rate, breathing rate etc., and provide information to the controller on the physiological characteristics whilst the user is exercising. Such sensors may be part of the exercise apparatus and/or standalone sensors, such as a breathing apparatus or chest/wrist-based heart rate monitor. The data obtained from these sensors may be obtained and logged by the controller, used to control the exercise apparatus, used to control the sensory stimulus output via the output unit 106, or used in other analysis, such as that described below.

[0054] The communication links 110, 112, and 114 may be wired or wireless and are each shown as two-way, but they may also be one-way. For example, the communication link 112 may be one-way from the controller 104 to the output unit 106. Likewise, the sensor may have a one-way communication link 114 from the sensor to the controller 106. With respect to the communication link 110, this link may be one-way from the exercise apparatus to the controller if the exercise apparatus is not a smart exercise apparatus able to be controlled be the controller 106 and may be two-way communication link when the exercise apparatus is a smart exercise apparatus. With respect to the wireless communication links, any wireless technology suited to the data transfer requirements and functionality of the various elements of the system may be used, such as Bluetooth, ANT+, or Wi-Fi for example.

[0055] Virtual reality-based exercise provides greater freedom when controlling user exercise since virtual reality-based exercise is not dependent on the available real-world environment and exercise apparatus parameters can be controlled to create scenarios that may not be possible in the real-world. Furthermore, exercise apparatus parameters may be more accurately controlled during virtual reality-based exercise compared to their real-world equivalent.

[0056] This additional level of control can allow influences on a user’s exercise performance to be investigated in further detail and with improved accuracy compared to performing such investigation during real-world exercise, since certain parameters may be independently controlled in a manner not practical in the real-world. For example, with respect to virtual-reality cycling, the gradient that a user climbs/descends and thus the resistance they experience through the exercise apparatus can be precisely controlled without the need to locate a particular real-world hill.

[0057] However, this increased level of control is not only limited to exercise apparatus parameters and their real-world equivalents but also applies to the sensory stimulus which is output to a user. For example, the virtual world presented to the user through the output unit can be precisely controlled in a manner not possible in the real-world.

[0058] The present disclosure provides approaches that exploit the increased level of control that virtual reality exercise allows in order to provide more in-depth analysis of a user’s exercise performance, which in turns enables increased levels of information about a user and the mechanisms underlying their performance to be determined. Discordant Stimulus During Virtual Reality Exercise

[0059] In existing approaches to virtual reality exercise the parameters of the exercise apparatus are controlled based on the sensory stimulus being output to the user (or vice versa) and also the user’s exercise input, such that the user’s experience through the exercise apparatus consistently corresponds to the sensory stimulus and expected real- world behaviour. For example, with respect to video-based virtual reality cycling, the resistance provided through the pedals of the bike is intended to accurately corresponds to the terrain and other parameters of the virtual world displayed by the video so as to replicate the equivalent real-world experience as closely as possible. In particular, the pedal resistance will be a function of one or more of the gradient of the terrain, the type of terrain, the speed of the rider, the wind speed, the type of bike, and a weight of the rider etc. Although such an approach contributes towards providing a realistic virtual reality exercise experience, it limits the information that may be obtained on the user and the mechanisms underlying their exercise performance and other physiological/psychological behaviours of a user, such as breathlessness or anxiety for example.

[0060] In accordance with the present disclosure, one or more sensory stimulus output to the user are discordant with (i.e. not in agreement with) one or more corresponding parameters of the exercise apparatus so that the influence of a user’s sensory input on their exercise performance or other physiological/psychological characteristics can be investigated and/or their exercise performance or other physiological/psychological behaviours manipulated via their sensory input. For example, with respect to video-based virtual reality cycling, the resistance provided through the pedals of a bike would be controlled so as not to correspond to the expected resistance given the steepness of the hill that is displayed to the user through the video. In other words, the simulated gradient (i.e. the resistance of the pedals) is discordant with the displayed gradient (i.e. the displayed steepness of the hill), such that there is an element of disassociation of a user’s expectation from their physical effort. With respect to running, the simulated gradient (i.e. the angle of the treadmill) is discordant with the displayed gradient (i.e. the displayed steepness of the hill). Alternatively, the speed displayed to the user may be different to the actual speed the user is running at. More generally, it may be defined that corresponding parameters of the sensory stimulus and the exercise apparatus are discordant with one another.

[0061] During the provision of the discordant stimulus, one or more physiological measurements of the user are obtained in order to monitor the performance of the user. For example, one or more physiological characteristics such as breathlessness of the user (measured automatically or via user self-reporting), a breathing rate (i.e. respiratory rate) of the user, a tidal volume of the user, a minute ventilation of the user, gas measurements of the user (e.g. expired oxygen and carbon dioxide), other metabolism-based metrics, a heart rate of the user, and a power output by the user through the exercise apparatus may be measured. Further physiological measurements may be taken dependent on the type of exercise, for example, with respect to cycling and running the cadence of the user may be measured, or the gait of the user may be measured with respect to running/walking. The physiological measurements may be obtained via automated sensor or obtained manually, where, as described above, the sensors may be separate sensors or included in one of the other entities of the virtual reality exercise system. Yet further physiological measurements/characteristics may include sweating, hunger, energy levels, discomfort, pain, and fatigue experienced by a user or a particular part of their body (e.g. legs with respect to cycling). Examples of physiological characteristics may include those which are subjective and thus change from user to user; however, this may be of a lesser concern if the data is not being compared to other users. Alternatively, particular measurement frameworks may be implemented to provide a more objective measure of subjective characteristics.

[0062] Once the physiological measurements have been obtained, based on these measurements and the discordant relationship between the parameters of the exercise apparatus and the sensory stimulus, an effect of the parameter of the stimulus on the performance of the user can be determined. For example, if the exercise is cycling and the parameter of the sensory stimulus includes a displayed gradient that is discordant with the simulated gradient of the exercise apparatus, the effect of the displayed gradient (i.e. user perception of the difficulty of the cycling) on the user’s exercise performance or other measured behaviour may be established.

[0063] Alternatively, if a relationship between a sensory stimulus and a user’s exercise performance has previously been established, the details of the relationship may be used to control the sensory stimulus in a discordant manner in order to manipulate their exercise performance and/or physiological symptoms, such as perception of breathlessness for example, which in turn may be used to treat breathlessness, potentially in a personalised manner. In other words, a user’s physiological symptoms may be manipulated via their expectation mechanisms.

[0064] Advantageously, such an approach is able to identify causes of, or contributing factors towards particular physiological symptoms that may not be easily identified via objective markers through conventional means, such as lung function tests for instance. The effect of a user’s expectation on their exercise performance or physiological symptoms (e.g. breathlessness) may also be determined without the use of more expensive and complex techniques such as functional magnetic resonance imaging (FMRI) for example. [0065] Although corresponding parameters of the sensor stimulus and the exercise apparatus are predominantly considered in this disclosure, non-corresponding parameters may be used, or a sensory parameter may be controlled independently of any exercise apparatus parameter. For example, a sound output or a visible weather condition may be controlled, where these do not have a corresponding or directly corresponding exercise apparatus parameter. With respect to a sound output, breathing sounds may be provided to the user, which do not have a directly corresponding exercise parameter but nevertheless may have an effect on the user’s perception of the effort they are exerting. However, it is anticipated that the use of corresponding sensory stimulus and exercise apparatus parameters provides an improved approach since discordant changes in the parameters may be more easily disguised to the user, thus reducing the impact of a user’s conscious decision/bias on their performance, which in turn assists with highlighting factors that affect their exercise performance or other physiological/psychological behaviours.

[0066] Although physiological measurements are primarily considered, the present disclosure is not limited to these, and measurements of psychological characteristics or characteristics that are a mix of psychological and psychological may obtained and analysed, such as a user’s anxiety levels. In particular, through the provision of discordant stimuli in an exercise-based scenario such that a user’s perceived effort and their actual effort can be monitored/controlled, the impact of physical inputs and psychological inputs on psychological aspects of the user can be investigated and determined. With respect to an example of a user’s anxiety levels (which may contribute to other symptoms such as breathlessness), the impact of their perception of the difficulty of the exercise (i.e. the displayed gradient for cycling) can be established by obtaining data on their anxiety level whilst discordant stimuli are applied. Put simply, in an extreme example, if there is only a correlation between a displayed gradient and a measured characteristics (i.e. the actual physical intensity of the exercise does not impact the measured characteristic), the mechanism behind the characteristics may be determined to be purely psychological. However, in reality, it is likely to be mix of mechanisms. Indeed, the disclosed technique may be particularly useful in determining the relative contribution of physical and psychological factors to a measured characteristic.

[0067] Figure 2 provides a flow diagram 200 illustrating an example method in accordance with the present disclosure for controlling a virtual reality exercise system such as that described with reference to Figure 1 .

[0068] At step 202, a sensory stimulus is output is through an output unit for presentation to a user of the virtual reality exercise system. The sensory stimulus may take any suitable form given the type of exercise apparatus, such as video/audio content, exercise apparatus position, vibration, haptic feedback, and air flow. The sensory stimulus may also include one or more of these forms. The sensory stimulus is determined by the controller and appropriate control signals then output to the output unit.

[0069] The sensory stimulus includes at least one displayed/observable parameter that corresponds (i.e. is related to) to a parameter of the exercise apparatus (i.e. simulated parameter). For example, with respect to cycling or running, when the sensory stimulus is video content, the sensory stimulus may include a terrain gradient in the form of a sloping road, where the terrain gradient corresponds to a pedalling resistance of the bike or the angle of the treadmill for running. When a plurality of different sensory stimuli are provided, more than one may have a parameter that corresponds to a parameter of the exercise apparatus. For example, with respect to cycling, a sensory stimulus may also include a tilting of the bike, where the angle of the tilting also corresponds to the terrain gradient. The sensory stimulus may also include one or more other features/parameters that are unrelated to a parameter of the exercise apparatus in order to enhance the virtual reality experience and/or manipulate the user’s exercise performance.

[0070] At step 204, a parameter of the exercise apparatus (i.e. an exercise apparatus parameter) is controlled according to a discordant relationship between the exercise apparatus parameter and the sensory stimulus. In particular, the exercise apparatus parameter is controlled by the controller based on a discordant relationship with a corresponding observable parameter of the sensory stimulus. For example, with respect to cycling, a simulated gradient (i.e. pedalling resistance) of the bike may be controlled in a discordant manner with respect to the steepness of the hill displayed to the user.

[0071] In this context, a discordant relationship is broadly defined as there being a difference between the exercise apparatus parameter and the corresponding observable parameter of the sensory stimulus. With respect to video-based virtual reality cycling, the observable parameter may be the gradient of the road (x%), and the exercise apparatus parameter a percentage gradient (y%) simulated via the resistance of the pedals. A discordant relationship may then be defined as x y for at least some of the period under consideration. The specific relationship between x and y may also be inconsistent (i.e. an inconsistent or variable discordant relationship) such that the is not a fixed relationship between x and y. For example, when a first observable gradient is 6% the resistance of the pedals will be correspond to that expected if the user were cycling up a real-world gradient of 6%; when a second observable gradient of 6%, the resistance of the pedals will be correspond to that expected if the user were cycling up a real-world gradient of 4%; and when a third observable gradient is 6% the resistance of the pedals will be correspond to that expected if the user were cycling up a real-world gradient of 10%. In other words, a discordant relationship may define a relationship where the exercise apparatus parameter does not correspond to that expected in the real-world for a corresponding observable parameter, or a relationship where the relationship between the exercise apparatus parameter and the observable parameter is not consistent. Although discordant, there may still be limiting or boundary conditions on the discordant relationship or limits on the parameter values, for example, there may be limits on the relative differences of the two parameters or the step changes that may occur so that the discordant relationship is not obviously apparent or at least disguised to some extent to the user. However, it is not essential that the user is unaware of the discordant relationship or that the discordant relationship is disguised. Furthermore, the specific discordant relationship and/or range/limits of the exercise apparatus parameters and observable parameter may be dependent on characteristics of the user or the intention of the procedure, for example, dependent on whether the intention is to improve exercise performance or to discover/investigate particular user characteristics or behaviours. With respect to limits on the individual parameter values, they may be set based on characteristics of the user, such as the fitness of the user for example.

[0072] Figure 3 provides an illustration of a discordant relationship between an observable gradient and a gradient simulated via the resistance of the pedals for a virtual reality cycling system (i.e. a parameter of the exercise apparatus). However, it may equally apply to other parameters of a sensory stimulus and exercise apparatus. Line 302 illustrates the gradient simulated via the resistance of the pedals and 304 illustrates the observable gradient of the road displayed as part of a video sensory stimulus (i.e. the parameter of the sensory stimulus). As can be seen, a sequence of different observable gradients are output as the sensory stimulus and the observable gradient does not correspond to the simulated gradient for the majority of the time. However, the discordant relationship may be presented for a minority of the time or for equal periods. More generally, the relative proportions of the simulation having discordant and non-discordant stimuli may vary and be controlled to more effectively manipulate the behaviour of the user. Furthermore, there is an inconsistent relationship between the observable gradient and the simulated gradient such that there is not a one to one mapping between an observable gradient and simulated gradient. As can be seen, the simulated gradient varies between being lower, equal to, and higher than the gradient displayed through the output unit. Although the changes in simulated and displayed gradients change according to a predefined period and are synchronised, the changes are not limited to this and may be non-period and/or non-synchronised. The specific gradients and differences in gradients that are simulated and displayed are also not limited to the values shown in Figure 3 but may take any suitable form. For example, the gradient values and/or differences in gradient values may be varied according to a predetermined pattern and/or randomly. [0073] The specific gradients shown in Figure 3 and also those used in the disclosed method may not be consistent with real-world gradients but should preferably be internally consistent. For example, when other parameters are constant, different periods of a 6% simulated gradient should result in the same resistance being presented through the pedals, although this resistance may not correspond to the resistance that would be experienced in the real-world by the user on a 6% gradient. Likewise, the 4% simulated gradient may not correspond to the resistance that would be experienced in the real-world by the user on a 4% gradient but the resistance presented through the pedals for the 4% gradient should be consistent and less than that presented for the 6% gradient.

[0074] At step 206, a user exercise input is detected though the exercise apparatus and measurements of one or more physiological characteristics of the user are obtained. The measurements of physiological characteristics may be detected via the exercise apparatus or separate measurement instruments as described with reference to Figure 1. The physiological characteristics that may be measured can take any suitable form depending on the nature of the relationship that is to be determined in step 208 and/or the nature of the feedback that may be used to control or update the sensory stimulus. Physiological measurements may include but are not limited to one or more of a breathlessness of the user, a breathing rate of the user, a tidal volume of the user, a minute ventilation of the user, a heart rate of the user, and a power output by the user through the exercise apparatus. The measurements may be automatically obtained via an automated sensor or manually obtained. For example, a breathless of a user may be based on the exercising user providing a relative breathlessness on an arbitrary scale (e.g. 0-10) at predetermined time intervals or at predetermined points/times during the exercise. The specific characteristic(s) measured may be dependent on the intention of the procedure, for example, if the intention to improve an athlete’s performance, their power output may be the most relevant physiological characteristic, whereas for a user experiencing unwanted breathlessness, their relative breathlessness may be more relevant.

[0075] Following step 206, the sensory stimulus may optionally be updated based on the user exercise input at step 210 so that the sensory stimulus responds to the user exercise input. For example, with respect to video content for cycling and running, the user will be seen to be moving through the displayed landscape. The information used to update the sensory stimulus may also include any of the obtained physiological measurements. In some implementations, the sensory stimulus and or the discordant relationship is updated based at least partially on the determined relationship so that the sensory stimulus and exercise apparatus parameter(s) can be optimised for the collection of the physiological measurements [0076] At step 208, a relationship between the sensory stimulus and one or more physiological characteristics for which measurements have been obtained is determined based on the based on physiological measurements and the discordant relationship between the exercise apparatus parameter and the sensory stimulus. The physiological measurements and the discordant relationship may be analysed to evaluate any potential relationship between the sensory stimulus and one or more of the physiological measurements. For example, a correlation coefficient between a physiological characteristic and a parameter of the sensory stimulus may be calculated in order to determine the effect that different sensory stimulus parameters have on a particular physiological characteristic. The appropriate choice of the discordant sensory stimulus enables specific relationships to be targeted. For example, using the discordant relationship illustrated in Figure 3 a relationship between a displayed gradient and a physiological characteristic can be effectively isolated.

[0077] Although not shown in Figure 2, the sensory stimulus and the parameter of the exercise apparatus may also be at least partially controlled based on the physiological measurements and/or the determined relationship, such that the sensory stimulus, the exercise apparatus and their discordant relationship can be controlled in real-time based on the user’s performance. In turn this may allow real-time adaptation of the data collection process in order to target sensory stimulus and/or exercise parameter values that result in more relevant or useful data for determining a relationship between the sensory stimulus and the physiological characteristic of the user. For example, if a significant correlation between a sensory stimulus parameter and physiological parameter is present only at specific values of the sensory stimulus or exercise apparatus parameters, the future values of the sensory stimulus or exercise apparatus parameters can be limited to such values.

[0078] The steps 202, 204, 206, and 210 of Figure 2 provide a sequence of discordant displayed and simulated parameters as shown in Figure 3, and thus may be considered to provide a first pair of discordant displayed and simulated parameter values during a first time period, a second pair of discordant displayed and simulated parameter values during a first time period, and so forth, where the pairs of parameters values are defined by a discordant relationship, such that there is no fixed relationship between the values of the two parameters (although there may be certain limits as discussed above). The measurements of a physiological characteristic obtained during each of these time periods can then be used to determine a relationship between the displayed parameter (i.e. a parameter of the sensory stimulus) and the physiological characteristic.

[0079] Figure 4a is a graph providing data points obtained via the provision of discordant stimulus in accordance with Figures 2 and 3, where the physiological characteristic is the power output through the pedals of a bike in virtual reality cycling system (i.e. the user’s power output), and the relevant parameter of the discordant sensory stimulus is the displayed gradient i.e. the observable slope of the road. The displayed gradient varies between 0% and 10% and the corresponding parameter of the exercise apparatus is the simulated gradient (i.e. resistance) and varies between 4% and 6%, where the resistance presented through the pedals for the simulated gradients was not calibrated for each user’s weight such that the resistance presented though the pedals for each simulated gradient is constant across the participating users. The virtual reality cycling system was based on Tacx Flow Smart turbo trainer with the user wearing a head-mounted display.

[0080] The data points were provided by 19 healthy participants (9 male, 10 female, median age 20 years; range 19-41 years) completing 14x100m work blocks with displayed gradients of 0%, 2%, 4%, 6%, 8%, 10% or 12%. Participants completed each displayed gradient twice, once with a simulated gradient (i.e. resistance) of 4% and once with a simulated gradient of 6%. In between each work block, participants completed a 30m recovery ride on a flat surface (0% displayed gradient) with a simulated gradient of 1 %. In total, participants completed 28 blocks. Participants cycled at a self-selected pace and were encouraged to maintain a cadence of approximately 80 revolutions per minute (RPM).

[0081] The graph has been generated using the function aov(power ~ slope_resistance * observed_slope, data = data_no_res_2) from RStudio version 1.3.1093, where the grey shaded areas represent a 95% confidence interval associated with each of the lines of best fit. As can be seen from Figure 4a, the line of best fit for simulated gradients of 4% and 6% each show a clear positive correlation between the displayed gradient and the power output by the user (i.e. an upwards slope from left to right), such that users, on average, output approximately 10w more power when the displayed resistance was 10% compared to 0% when the other parameters were constant. Furthermore, there is a clear gap between the two lines of best fit, showing that there is a positive correlation between the simulated gradient and the power output by the user, such that users, on average, output approximately 20w more power when the simulated gradient was 6% compared to 4%. Consequently, it can be concluded that the power output by the user correlates not only with simulated gradient (i.e. the resistance experienced through the pedals) but also correlates and thus has an independent relationship with the displayed gradient. Hence showing that the power output by the user is dependent on the perceived gradient as well as the actual gradient experienced by the user. In other words, the power output by the user is partially dependent on a gradient that they observe through the sensory stimulus.

[0082] Figure 4b is a graph providing an enlarged version of the lines of best fit and the 95% confidence intervals (grey shaded area) that are illustrated in Figure 4a. Referring to Figure 4b, the positive correlation between the displayed gradient and the power output by the user is clearly illustrated, as well as the difference in power output between users when cycling on a 4% and a 6% simulated gradient.

[0083] Figure 5a provides an estimated correlation coefficient 502 between a displayed gradient and breathlessness of a user along with their standard errors based on the provision of discordant displayed and simulated gradients in a video-based virtual reality cycling system. An estimated coefficient 504 between a power output by a user and their breathlessness is also provided for reference. The coefficients have been calculated in accordance with the function lmer(breathlessness ~ S_power + S_observed_slope (1\Participant), data = data_no_res_2, REML = F, from the Ime4 package of RStudio version 1.3.1093. The breathlessness of the user in this example has been obtained via an exercising user verbally providing a relative measurement of their breathless on a scale of 1-10; however, other appropriate measures, scales or approaches to providing the measurements (e.g. buttons, touchscreens etc.) may also be used. As can be seen the displayed gradient coefficient reflects that there is a significant positive correlation between user breathlessness and displayed gradient. Consequently, this indicates that during exercise breathlessness may be reduced (increased) by providing an observable gradient that is less than (or greater than) the simulated gradient. As one would expect, there is a strong correlation between a user’s power output and breathlessness. In particular, the coefficients of Figure 5a indicate that approximately 70% of the breathlessness reported relates to physical effort (i.e. output power) and 30% relates to the displayed gradient. In turn, this means that a user may exercise at a higher intensity with a same level of breathlessness via the display of discordant stimuli.

[0084] Figure 5b provides estimated correlations coefficients of a displayed gradient 510, an age of the users 512, and a BMI of the users 514 with a power output of the users along with their standard errors based on the provision of discordant displayed and simulated gradients in a video-based virtual reality cycling system as described with reference to Figures 3, 4a, 4b, and 5a.

[0085] As can be seen from Figure 5b, there is a relatively small positive correlation between the displayed gradient and the output power, which corresponds to that shown in Figures 4a and 4b. With respect to age and BMI, they have a negative and positive correlation with output power respectively, indicating that age and BMI can be used as predictors of power. Given that age and BMI would generally be expected to have negative and positive correlations respectively in adults cycling, this supports the validity of the experimental setup.

[0086] Although correlation coefficients are illustrated in Figures 5a and 5b, alternative outputs may be obtained from the disclosed method. For example, the collected data may be used to determine parameters of any appropriate model that expresses the relationship between the various sensory stimulus and physiological parameters.

[0087] As noted with respect to Figure 3, the values and/or differences in parameter values may be varied according to a predetermined pattern and/or randomly, so that parameter values and their differences can be varied so that targeted physiological measurements can be obtained. For example, different ranges of values and differences between the parameters can be used so that the relationship between a parameter of the sensory stimulus and a physiological characteristic at those ranges of values and differences between the parameters can be obtained. Likewise, although the relationships shown in Figures 4 and 5 are based on displayed gradient values between 0% and 10% and the simulated gradient varies between 4% and 6%, more detailed relationship information may be obtained by generating separate graphs/correlation coefficients when the parameters values and their differences are of a certain value.

[0088] Figures 6a, 6b, 6c, and 6d illustrate parameters and results from a further example implementation of the method illustrated by and described with reference to Figure 2. In this further implementation leg discomfort resulting from cycling was also measured in addition to breathlessness. More specifically, twenty participants (i.e. users) were recruited to a random order controlled, double blinded study. Participants attended a human exercise laboratory on two occasions to undertake virtual reality stationary cycling, separated by approximately 1 week. At one visit, the virtual reality hills displayed during cycling were concordant with the resistance applied to the pedals (that is, when a very steep virtual reality hill was observed, the resistance applied to the pedals by the cycling ergometer was appropriately high). At the other visit, the virtual reality hills displayed during cycling were discordant with the resistance applied to the pedals (that is, the hills were less steep than might be expected given the resistance applied to the pedals by the cycling ergometer). Participants completed two exercise bouts at each visit. The order in which participants completed concordant and discordant visits was randomised, as was the order of the different slopes within each session. Each session included a virtual course with 6 x 300 m hills, separated by 100 m flat sections (flat sections not shown in Figures 6a and 6b), and the cycling resistance (that is, the resistance applied by the bike to turn the pedals) varied throughout each exercise bout but was the same for all hills across all exercise bouts (that is, resistance was not an experimental variable). At the end of each hill, participants were asked to rate their breathlessness (1 nothing at all - 10 most breathless imaginable) and leg discomfort (1 nothing at all - 10 most leg discomfort imaginable). By dissociating effort expectation (that is, the expectation of how difficult/tiring it will be to cycle up a hill based on its gradient) from actual effort (the amount of power (W) required to turn the pedals), it is possible to test the independent effects of manipulating effort expectation on the subjective sensations of breathlessness and leg discomfort; two perceptions that are known to limit an individual’s exercise capacity.

[0089] Figure 6a provides an illustration of an example concordant relationship between an observable gradient and a gradient simulated via the resistance of the pedals for a virtual reality cycling system (i.e. a parameter of the exercise apparatus) used when the virtual reality hills displayed during cycling were concordant with the resistance applied to the pedals according to the further implementation. Line 602 illustrates the gradient simulated via the resistance of the pedals and 604 illustrates the displayed gradient of the road displayed as part of a video sensory stimulus (i.e. the parameter of the sensory stimulus).

[0090] Figure 6b, in a similar manner to Figure 3, provides an illustration of an example discordant relationship between an observable gradient and a gradient simulated via the resistance of the pedals for a virtual reality cycling system (i.e. a parameter of the exercise apparatus), used when the virtual reality hills displayed during cycling were discordant with the resistance applied to the pedals according to the further implementation. Line 612 illustrates the gradient simulated via the resistance of the pedals and 614 illustrates the displayed gradient of the road displayed as part of a video sensory stimulus (i.e. the parameter of the sensory stimulus).

[0091] As described with reference to Figure 3, the specific gradients shown in Figures 6a and 6b may not be consistent with real-world gradients but should preferably be internally consistent.

[0092] Figure 6c provides an estimated correlation coefficient between breathlessness of a user and a time elapsed 620, between breathlessness of a user and a displayed gradient 622, and between breathlessness of a user and a power output by the user 624 based on the provision of discordant displayed and simulated gradients in a video-based virtual reality cycling system as described with reference to Figures 2, 6a and 6b and the further example implementation. The coefficients have been calculated by fitting a linear mixed model to predict perception of breathlessness (1-10 scale) with the power required to cycle (W), the displayed gradient of the virtual reality hills (%) and time elapsed (s) as fixed effects, with participant included as a random effect. The model explained 67% of variation in perceptions of breathlessness during cycling. The fixed effects alone (displayed gradient, power and time elapsed) explained 38% of variation in perceptions of breathlessness. The linear mixed model was based on the Ime4 package of RStudio version 2023.03.0+386. The breathlessness of the user in this example has been obtained via an exercising user verbally providing a relative measurement of their breathlessness on a scale of 1-10; however, other appropriate measures, scales or approaches to providing the measurements (e.g. buttons, touchscreens etc.) may also be used. [0093] With respect to the estimated correlation coefficient between breathlessness of a user and a time elapsed 620, the effect of time elapsed is statistically significant and positive (beta = 2.94e-03, 95% Confidence Interval [2.86e-03, 3.01 e-03], t-statistic (17848) = 80.49, p-value < .001).

[0094] With respect to the estimated correlation coefficient between breathlessness of a user and a displayed gradient 622, the effect of displayed gradient is statistically significant and positive (beta = 0.34, 95% Confidence Interval [0.32, 0.36], t-statistic (17848) = 31.45, p-value < .001).

[0095] With respect to the estimated correlation coefficient between breathlessness of a user and a power output by the user 624, the effect of power is statistically significant and positive (beta = 0.79, 95% Confidence Interval [0.77, 0.82], t-statistic (17848) = 58.66, p- value < .001).

[0096] As can be seen, the displayed gradient coefficient reflects that there is a significant positive correlation between user breathlessness and displayed gradient. In other words, as shown with respect to Figures 5a and 5b, this further experimentally obtained data confirms that the steepness of the observed virtual reality hill gradient had a significant and independent effect on individuals’ perception of breathlessness. Consequently, this indicates that during exercise breathlessness may be reduced (increased) by providing an observable gradient that is less than (or greater than) the simulated gradient. As one would expect, there is a strong correlation between a user’s power output and breathlessness. In turn, this means that a user may exercise at a higher intensity with a same level of breathlessness via the display of discordant stimuli.

[0097] Figure 6d provides an estimated correlation coefficient between leg discomfort of a user and a time elapsed 630, between leg discomfort of a user and a displayed gradient 632, and between leg discomfort of a user and a power output by the user 634 based on the provision of discordant displayed and simulated gradients in a video-based virtual reality cycling system as described with reference to Figures 2, 6a, and 6b and the further example implementation. The coefficients have been calculated by fitting a linear mixed model to predict perception of leg discomfort (1-10 scale) with the power required to cycle (W), the displayed gradient of the virtual reality hills (%) and time elapsed (s) as fixed effects, with participant included as a random effect. The model explained 66% of variation in perceptions of leg discomfort during cycling. The fixed effects alone (displayed gradient, power and time elapsed) explained 33% of variation in perceptions of leg discomfort. The linear mixed model was based on the Ime4 package of RStudio version 2023.03.0+386. The leg discomfort of the user in this example has been obtained via an exercising user verbally providing a relative measurement of their leg discomfort on a scale of 1-10; however, other appropriate measures, scales or approaches to providing the measurements (e.g. buttons, touchscreens etc.) may also be used.

[0098] With respect to the estimated correlation coefficient between leg discomfort of a user and a time elapsed 630, the effect of the time elapsed is statistically significant and positive (beta = 3.02e-03, 95% Confidence Interval [2.95e-03, 3.09e-03], t-statistic (17848) = 85.20, p-value < .001).

[0099] With respect to the estimated correlation coefficient between leg discomfort of a user and a displayed gradient 632, the effect of the displayed gradient is statistically significant and positive (beta = 0.24, 95% Confidence Interval [0.22, 0.26], t-statistic (17848) = 23.22, p-value < .001).

[00100] With respect to the estimated correlation coefficient between leg discomfort of a user and a power output by the user 634, the effect of power is statistically significant and positive (beta = 0.66, 95% Confidence Interval [0.64, 0.69], t-statistic (17848) = 50.29, p- value < .001).

[00101] As can be seen, the displayed gradient coefficient reflects that there is a significant positive correlation between user leg discomfort and displayed gradient. In other words, this experimentally obtained data shows that the steepness of the observed virtual reality hill gradient had a significant and independent effect on individuals’ perception of leg discomfort. Consequently, this indicates that leg discomfort during exercise may be reduced (increased) by providing an observable gradient that is less than (or greater than) the simulated gradient. As one would expect, there is a strong correlation between a user’s power output and leg discomfort. In turn, this means that a user may exercise at a higher intensity with a same level of leg discomfort via the display of discordant stimuli.

[00102] Overall, the results described with reference to Figures 5a, 5b, 6c, and 6d, show that users reported less breathlessness and leg discomfort for equivalent workload when the displayed gradient was shallower than expected for the resistance applied to pedalling. [00103] Although leg discomfort has been considered in relation to Figures 6a-6d, the applied method illustrated by Figure 2 may be applied to other physiological characteristics. [00104] As noted above, results such as those illustrated in Figures 4, 5a, 5b, 6c, and 6d that have been obtained by following the method of Figure 2 enable a relationship between sensory stimulus and physiological characteristics to be determined. These determined relationships may then be used to identify or exploit certain user performance traits, which may be utilised in, among others, athletic training and healthcare applications.

[00105] In particular, by identifying relationships between sensory stimulus and physiological characteristics during exercise, the relative influence of psychological influences on physical performance can be inferred. In turn, this enables the causes (i.e. physiological or psychological) of particular medical or physiological conditions and causes of user exercise performance limitations to be identified with improved accuracy and/or lower cost compared to approaches such as FMRI. Additionally, via the manipulation of the sensory stimulus presented to a user during exercise based on knowledge of a relationship between sensory stimulus and physiological characteristics, a user’s existing performance limitations may be altered so that they may train with increased intensity.

[00106] With respect to healthcare scenarios, if a significantly different (e.g. a greater or lesser) than usually expected correlation coefficient or model parameter is determined, or a relationship between a problematic physiological characteristics (i.e. symptoms such as high heart rate, leg discomfort/fatigue or high levels of breathlessness) and a sensory stimulus parameter is identified via the method described with reference to Figure 2, it can be inferred or diagnosed that psychological factors play a significant role in the causation of the problematic physiological characteristics. Consequently, treatment to address the problematic physiological characteristics can more accurately target the causes of the problematic physiological characteristics. For example, considering the issue of chronic breathlessness in a user, if there is a stronger than expected relationship between a sensory stimulus parameter and breathlessness it is an indicator that psychological factors are likely to be a significant contributing factor to their condition, and thus treatment can be focussed on psychological factors. Expected (i.e. normal) correlation coefficients or model parameters may be obtained from users that do not have any identifiable abnormal behaviours and/or may be obtained via analysis of a user’s performance when the sensory stimuli are nondiscordant.

[00107] With respect to athletic performance or general fitness scenarios, the extent to which there is a relationship between physiological characteristics and sensory stimulus enables more effective identification of the causes of performance limitations. For example, if there is significant relationship between a sensory stimulus parameter and a particular physiological characteristic that is presenting a barrier to performance, the particular psychological mechanism that plays a role in the sensory pathways may be targeted in order to remove or lessen the barrier. For example, with respect to cycling, if a maximum power output of a user has a significant correlation to the displayed gradient, it can be established that there is a psychological barrier related to the user’s perception of hill steepness influencing their maximum power output. With this knowledge, steps can be taken to address this psychological barrier and thus improve performance. Similarly, if leg discomfort has a significant correlation to the displayed gradient, it can be established that there is a psychological barrier related to the user’s perception of hill steepness influencing their how much discomfort they experience in their legs and the extent to which they are able to continue exercising at a particular intensity. For example, by the provision of discordant gradients to a cyclist with such a correlation, the cyclist may exercise at a higher intensity (e.g. output a higher power) for a longer period of time before their leg discomfort forces them to reduce their intensity (e.g. output power) or stop.

[00108] With respect to healthcare and athletic scenarios, if a particular relationship has been identified between a sensory stimulus parameter and a physiological characteristic, it may be exploited to enable the user to receive treatment/rehabilitation at a higher intensity or the user to train at a higher intensity. For example, if it is established via the method of Figure 2 that there is a significant relationship between a particular sensory stimulus parameter and a physiological characteristic that restricts training or treatment, the sensory stimulus parameter may be appropriately controlled in order to enable the level of the physiological characteristic to be controlled. Taking the example of breathlessness and cycling, if a user’s breathlessness is restricting exercise of a particular intensity required for medical or athletic purposes to be achieved and it has been established that there is a significant relationship between breathlessness the user experiences and a sensory stimulus of a displayed gradient, the displayed gradient may be appropriately controlled in order to increase the point at which a the user becomes breathless, thus enabling them to cycle with increased intensity. In an example clinical scenario, such an approach could be used to enhance pulmonary rehabilitation in patients with chronic respiratory diseases, or to enhance pre-surgical training (^rehabilitation ¹ ) in patients about to undergo major surgery. Consequently, in scenarios where a psychological component influences the relationship between the sensory stimulus and the physiological characteristic, the psychological component can be exploited or controlled in order to control the physiological characteristic of the user.

[00109] An alternative approach to interpreting the significance of a relationship between sensory stimulus and physiological characteristics during exercise is the concept of perceived effort versus actual effort. In particular, if there is significant relationship between a sensory stimulus and a particular physiological characteristic, it can be established that a user’s perception of a particular factor (e.g. effort they are making in the case of a displayed gradient) is significant in the behaviour of the physiological characteristics. In the case of breathlessness, it indicates that if a user expects to become breathless (e.g. due to the perceived steepness of a hill) they will become breathless more easily. In turn this means that there is scope for disassociating a user’s perception of effort from their actual effort. Consequently, information on the role of a user’s perception in their physiological behaviour can be therefore identified, and the information used in the various healthcare and athletic scenarios discussed above.

[00110] Although the examples above have concentrated on scenarios where a single sensory stimulus parameter and a single physiological characteristic are considered, the disclosed approach is not so limited, such that a plurality of sensory stimulus may be controlled and/or a plurality of physiological characteristics may be intended to be manipulated. For example, it may be established via the approach of Figure 2 that a combination of stimuli have a cumulative relationship and thus by combining a number of discordant stimuli a larger effect on a physiological characteristic can be achieved. Alternatively, two different physiological parameters may be simultaneously but independently manipulated via two different sensory stimuli during an exercise session.

[00111] Figure 7 provides a flow diagram 700 illustrating an example method in accordance with the present disclosure for controlling a virtual reality exercise system such as that described with reference to Figure 1 , where a sensory stimulus that is discordant with a corresponding exercise apparatus parameter is presented to a user. However, in contrast to the flow chart 200 of Figure 2, the flow chart 700 illustrates a method for exploiting the presence of a relationship between a sensory stimulus and a particular physiological characteristic using discordant stimulus in order to manipulate exercise performance. In other words, the provision of discordant sensory stimulus and exercise apparatus parameters may be used to suppress particular physiological characteristics.

[00112] At step 702, a sensory stimulus is output via the output unit under control of the controller. The sensory stimulus may take the form of any of those previously discussed, such as video content for example. The sensory stimulus also includes at last one parameter that corresponds to a parameter of the exercise apparatus.

[00113] At step 704 a parameter of the exercise apparatus is controlled by the controller based on a discordant relationship between a parameter of the sensory stimulus and a corresponding parameter of the exercise apparatus, where the exercise apparatus may be any of those previously discussed. With respect to cycling or running, the discordant relationship may define that a gradient simulated by the exercise apparatus is either less than or greater than the displayed gradient. If the intention is to reduce a user’s perceived effort compared to their actual effort, the simulated gradient may be higher than the displayed gradient. The relationship between the parameter of the sensory stimulus and exercise apparatus parameter may also be discordant and inconsistent, such as the relationship illustrated in Figure 3. The specific relationship between the sensory stimulus and the exercise apparatus parameter may be based on a relationship determined using the method of Figure 2 for the user or a generalised relationship that has been determined across a set of users that may or may not have included the present user. For example, if via the approach of Figure 2 a correlation between a displayed gradient and breathlessness predominantly occurs when the difference between the displayed gradient and the simulated gradient is small, the differences between the displayed gradient and simulated gradient may be maintained at a small value so that breathlessness can be effectively controlled via the displayed gradient. [00114] At step 706 the user exercise input through the exercise apparatus is detected. For example, with respect to cycling the user’s turning of the pedals and the associated power may be detected. However, any appropriate input for the type of exercise apparatus may be detected.

[00115] At step 708 the sensory stimulus is updated based on user exercise input and the updated sensory stimulus output to the user in order to provide a reactive virtual reality exercise experience. For example, a video may be updated to simulate the user travelling along a road or path at a speed determined according to their pedalling power.

[00116] In some examples, the approaches of Figures 2 and 7 may be combined such that the discordant relationship between sensory stimulus and the parameter of the exercise is optimised whilst the user is exercising. For instance, physiological measurements may be obtained whilst the user is exercising according to Figure 7 (e.g. at step 706) and a relationship between parameters of the sensory stimulus and exercise apparatus determined. This determined relationship may then be used to optimise the discordant relationship between sensory stimulus and the parameter of the exercise used in step 704. Such an approach allows the method of Figure 7 to adapt in real-time to the performance of the user and thus optimise the exercise being performed by the user.

[00117] Throughout the description and claims of this specification, the words “comprise” and “contain” and variations of them mean “including but not limited to”, and they are not intended to (and do not) exclude other moieties, additives, components, integers or steps. Throughout the description and claims of this specification, the singular encompasses the plural unless the context otherwise requires. In particular, where the indefinite article is used, the specification is to be understood as contemplating plurality as well as singularity, unless the context requires otherwise.

[00118] Features, integers, characteristics, compounds, chemical moieties or groups described in conjunction with a particular aspect, embodiment or example of the invention are to be understood to be applicable to any other aspect, embodiment or example described herein unless incompatible therewith. All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and/or all of the steps of any method or process so disclosed, may be combined in any combination, except combinations where at least some of such features and/or steps are mutually exclusive. The invention is not restricted to the details of any foregoing embodiments. The invention extends to any novel one, or any novel combination, of the features disclosed in this specification (including any accompanying claims, abstract and drawings), or to any novel one, or any novel combination, of the steps of any method or process so disclosed.