[Proposal] How does the heart manipulate our perceptual experience?

The human brain takes in the information of its environment using the body’s sensory organs. This information comes in a very chaotic form and is unfit for our brains to create a conscious perception of it. To remedy this chaos, according to Bayesian theory, the brain makes predictions about what the incoming sensory information means, based on its understanding or ‘model’ of previous similar sensory information. Predictions that fit the incoming sensory information will increase the probability of a ‘model’ shaping perception and bad predictions serve as fuel for the creation of better models.

While external sensory information is used by our brains to create and revise predictive models about the world, internal physiological (bodily) states also influence the way in which these models are selected, and thereby shape our conscious experience [1] . Examples of internal interoceptive (bodily) signals affecting perception can be evidenced by the effects of the heart on experience of body ownership [2] and emotion perception [3] . When feedback on a person’s physiological state, such as a heart rate (HR), is felt to be higher or lower than their actual HR, a change in perception occurs. For example, emotionally neutral faces appear to have increased emotional intensity, suggesting that when there is ambiguity in our lives, interoceptive awareness (sensing the internal state of the body) drives conscious perception.

The importance of this research lies in the apparent decrease in interoceptive awareness in patients with distinct psychiatric disorders. Here, interoceptive problems are linked with the individual expression of psychiatric symptoms (e.g. [4; 5] ). Understanding the inner workings behind false physiological feedback on perception has implications for the way we investigate and treat interoceptive anomalies in psychiatric populations.

We will use an auditory feedback paradigm [3] to manipulate how a person feels their internal physiological state (HR). Changing the rate in which participants feel pulse-like sensations in their auditory canal, we will ask participants to rate the emotional intensity of morphed-from-neutral smiling and frowning face images, while presenting these pulse-like sensations higher and lower than their normal heart rate. Put succinctly, we will give auditory signals that ‘feel physiological’, to change the brain’s inner interoceptive ‘feeling’ of heart rate. We will monitor how this alters perception, using subjective judgments of emotional faces.

We hypothesise : Unattributed arousal (fast stimulation) will bias participants to enhanced emotional ratings for close-to neutral face stimuli.

Individual differences in affectivity (as measured by the PANAS scale) may lead to biasing toward negative emotional ratings.

Measures of stronger explicit (conscious) interoceptive representation (better heartbeat detection, better interoceptive conscious awareness & metacognition) will decrease impact of pulse-like stimulation on emotional perception bias.

There will be a difference in emotional ratings for each condition between experimental and control groups.

There will be no difference in demand characteristics within conditions of the control group.

Methodology: (Note: experiment paradigm will run through the Gorilla.sc experiment platform, the minimal costs of this service shall be directly funded by our team). Participants will be run through a pre-screening process upon responding to a study advertisement. After which, informed consent and basic demographics (reconfirmation of pre-screen questions) will be taken along with the following questionnaires: The Positive and Negative Affective Scale (PANAS), The Multidimensional Assessment for Interoceptive Awareness (MAIA) and the State-Trait Anxiety Index (STAI).

Participants will be split into (optional) control and experimental groups and will be directed to a screen in which they must measure and input their average heartrate by counting heartbeats over a one-minute period. After which, they will be directed to a pre-made condition tree that best represents their average HR (e.g. HR of 82 will put them in the 80-85 condition tree).

Consistent with previous studies [3] , they will then be exposed to an auditory stimulation that will simulate a heartbeat at either 1) above, 2) around 3) below their current heartbeat or 4) zero stimulation (4 conditions total) in a randomised order. This will be described as a pulse that will feel like your heartbeat. During this time, they will be asked to judge the emotional intensity of morphed-from-neutral faces using their keyboard, from -4 for a negative emotion to +4 for a positive emotion (with 0 being neutral/no emotion).

Controls will be told to pay less attention to the emotional faces and to make their emotional ratings based on the way they THINK they SHOULD response to PLEASE the researcher.

Conditions and number of trials are as follows:

Conditions (number of stimulation blocks): Below HR (16), around HR (16), Higher than HR (16), No auditory stimulation (16)

Length of auditory stimulation blocks (number of trials): 20 seconds (5)

Trials: 4 seconds each; face-presentation 1000ms, response-time 3000ms

Total time: ~21 minutes.

Justification of sample size and pricing

According to the equation provided by http://www.raosoft.com/samplesize.html:

For a study with 95% confidence level and 5% margin of error a total of 377 participants would need to be collected. Imaging data collected at our lab using a 3T fMRI scanner will total 40 participants, making it logical to increase the size to 400 for 10 to 1 scale. And additional (optional) 400 Participant would be collected as a control measure to assess the effect of demand characteristics on emotionality ratings between conditions. This would make a total of 800 participants.

Pricing: 400 experimental Participants = £2000 + 400 (optional) control participants = £2000

Total = £2000/£4000

Findings and analysis code: All findings and their interpretations within the current literature will be published in a peer-reviewed scientific journal pending acceptance. Any MATLAB and R scripts used to extract and analyse the data from Gorilla.sc’s outputs will be published in an online repository with the fully anonymised data. Links to open-source questionnaires discussed in methodology will also be provided for replication purposes. There is also the potential to upload the Gorilla paradigm onto their open materials repository (Open Materials).

Pre-registration: https://aspredicted.org/4ut52.pdf