Download the full Insight Project article here.Download
The Primacy of Fear
The project must begin somewhere, and we suggest that the experience casting the largest shadow is fear. It is present in anxiety, phobias, PTSD and psychosis amongst others. Moreover, when we think of experiences like voices or visions, it is often the accompanying fear rather than the perceptual experience that causes distress. Fear makes us lose grasp on reality, act irrationally and is at the root of much mental suffering.
The intuitive approach is that symptoms beget suffering and so curing the symptoms is key to curing the suffering. This is as intuitive as plucking a guitar string to make a sound. But what if we turn it on its head? Can making a sound vibrate the guitar string? Well yes, it can and does. Pioneering psychologist William James was one of the first to take this counter-intuitive but intriguing perspective when he said that “we feel sorry because we cry, angry because we strike, afraid because we tremble”. In other words, the emotions come from the physical responses, not the other way around. Though a simplification, this is partly true.
Can controlling fear itself help ease our symptoms? This is the core idea we will explore. If true, it offers the possibility of a focused interventional approach that could cut off suffering at its source:
Fear itself is the key experience we will strive to overcome to reduce mental suffering
Exposing and Identifying Fear
At its most basic, fear is a mental state with an associated physical response. The physical and mental are intimately linked. Indeed, it has been argued that one cannot experience fear in the absence of its physical attributes.
But can we “fingerprint” fear by examining its physical, measurable attributes? If we can then we are confronted with an exciting possibility: by identifying the parameters of fear, we can measure them and, critically, characterise how they change in relation to particular situations and interventions, both externally created by the skilled game designer and internally shaped by the individual.
The responsive and absorbing narratives and contexts of the video game become the arena within which the individual learns to recognise, identify and even harness their physical and physiological responses. The game becomes the learning environment in which they can gain control of fear.
These are exciting ideas with important implications but reality dictates that we begin with more fundamental and basic questions including:
- What are the core physical and physiological features of fear and what are the best ways to measure them?
- Do these physical features reliably indicate fear? Do they vary from person to person? Do they vary from one time to another within a single person?
- Fear seems to come in many forms: can we distinguish sub-types of fear subjectively and in terms of the accompanying physical and physiological responses? Are there a plethora of mental states of fear that we do not yet have words or categories for?
Biometric signals are not new and if there were to be easy answers to the above, they would have been found by now. The reality is that biometric signals present a great deal of noisy data that can be hard to interpret or make sense of. Without a precise context for what is going on in a subject’s mind, the data is next to useless. But here is why we shouldn’t give up:
- Low-level biometrics like pulse, sweat and breathing monitors are becoming commoditised, opening the door to a lot of data
- EEG and neurotechnology has advanced greatly, opening new data streams
- Big data analysis and machine learning are creating superhuman pattern-spotting algorithms
- New methods of analysis such as voice, eyes and facial recognition are bringing us higher-level signals to add to the mix. Furthermore, the precise and accurate quantification of these measures in naturalistic settings is becoming ever more achievable.
So, a strategy may be to look for a “fingerprint” of fear across multiple biometric data streams in a person rather than to try and derive a meaningful fear state from any one noisy data stream. It may well be that identifying these fingerprints requires machine learning, as they may be too obscure or impractical to identify manually.
We must reveal fear in its many guises from biometric data using big data analysis
Invoking Fear through Control Contexts
We cannot bring a scientific approach to the understanding of fear without the capacity to observe states of fear. Experimentally, this means that we have to be able to induce a controlled state of fear in an individual so that we can detect and study their responses in the many forms that they take.
In this respect, the game designer has a set of powerful skills. Games can induce precise fear states and we do so through storytelling and immersion. Take the opening of Hellblade in the canoe: one can observe that most gamers playing the game on Twitch share the experience of ‘dread’ and one can imagine that they would produce similar physiological signals as a result.
With the capacity of games to create precise simulated scenarios – control contexts – we can model ideal settings within which we can do unique experiments on fear, inducing states, recording signals and manipulating immersive contexts such as VR and non-VR settings.
Against these control contexts and suitable subjects, we will measure every signal we can and look for fingerprints in the noise. We can increase the sample size to many players and see if there is a common fingerprint for fear states across different people. We can also look for an overarching signal for fear itself.
Our approach explicitly recognises that key parameters may not be the absolute values of particular signals (e.g. heart rate) but rather the dynamic shifts in these values (e.g. beat-to-beat variability of the heart, or rate of change of respiration). Indeed, core signals may lie in higher level interactions between variables, such as the degree to which heart rate variability is determined by respiration patterns.
We will use videogame scenes to give control contexts to biometric signals to help phenotype fear