The tasks#
Cardioception provides two established tasks, Heart Rate Discrimination (HRD) and Heartbeat Counting (HBC), and two beta tasks, the Phase Adjustment Task (PAT) and Heartbeat Synchrony Discrimination (HSD). The beta tasks are close to feature complete, but validation is ongoing. Implementations of heartbeat discrimination and the method of constant stimuli are planned.
This page compares the tasks’ measures, procedures, and practical requirements. The individual task pages explain how to configure and run each one.
Task |
||||||
|---|---|---|---|---|---|---|
Measure |
Bias (threshold) and precision (slope) of judgements of one’s own heart rate, with confidence |
Heartbeats reported against heartbeats recorded over fixed windows, with confidence |
Consistency of the chosen tone delay or phase across trials, with confidence |
Where in the cardiac cycle the heartbeat is felt ( |
Discrimination of synchronous from delayed stimulus trains [Whitehead et al., 1977] |
Proportion of tones judged simultaneous at each delay [Brener et al., 1993] |
Reference |
Legrand et al. [2022] |
Brener et al. [1993] |
Whitehead et al. [1977] |
Brener et al. [1993] |
||
Adaptive |
Yes: a Psi staircase sets the tone rate |
No |
No: the participant adjusts the delay |
Yes: Bayesian placement of the tone delay |
No |
No |
Hardware |
Pulse oximeter, or ECG or PPG over LSL; headphones; mouse or keyboard |
Pulse oximeter, or ECG or PPG over LSL; keyboard |
ECG over LSL; sound card; mouse wheel; a tactor for the control task |
ECG over LSL; sound card; keyboard |
||
Trials and minutes |
120 trials, about 20 minutes plus the tutorial |
6 counting windows, about 4 minutes |
28 trials and 8 practice, about 30 minutes |
15 to 60 trials; 60 took 7 minutes |
||
Status |
Established |
Established |
Beta |
Beta |
Coming soon |
Coming soon |
Every task writes to the same run directory layout, with the trial table in
*_final.txt; Running a session
describes it.
Strengths and limitations#
The tasks differ in what they ask participants to judge and what can be inferred from their responses.
The HRD. It separates the bias of heart rate judgements from their precision, includes a within-person exteroceptive comparison, adapts the stimulus with a staircase, and runs on a pulse oximeter. Hierarchical models with normative priors are available for its data. It measures judgements of heart rate, which combine beliefs about one’s heart rate with whatever is sensed in the listening window. It does not show whether or when a person feels individual heartbeats; see what you may conclude. A full session takes about 20 minutes plus the tutorial.
The HBC. It is brief, can run with a pulse oximeter, and supports comparisons with a large existing literature. A plausible estimate of heart rate can produce a good counting score without detecting any beats [Windmann et al., 1999, Brener and Ring, 2016], and counting scores are unrelated to heartbeat detection [Ring and Brener, 2018]. A counting score alone therefore provides limited evidence of cardiac sensitivity.
The PAT. It connects to the phase adjustment literature and classifies participants against chance with a fixed number of trials. It cannot separate where the sensation lies from how precise it is, and a participant can be consistent through a habit of search rather than a sensation; the PAT and the HSD sets the two side by side. This port is in beta, and its timing differs from the hardware original.
The HSD. It estimates the location and temporal precision of heartbeat
sensations within the cardiac cycle. It adapts the multi-interval design of the method of
constant stimuli, which takes about 35 minutes [Brener and Ring, 2016], and
selects delays according to their expected information gain. It is in beta: not yet
validated, unable to probe delays below 130 ms, and sensitive to jitter in a
Bluetooth ECG, which widens sigma. Read the design page
before using it.
The PAT and the HSD#
Both tasks time a tone from the R-wave, on a delay in milliseconds or a phase of the cardiac cycle, and both ask for a confidence rating. They differ in the psychophysical method.
PAT |
HSD |
|
|---|---|---|
Method |
Adjustment: the participant turns a dial until the tones feel synchronous, then presses space |
Constant stimuli made adaptive: the procedure sets the delay, the participant answers yes or no |
Who samples the delays |
The participant, freely, for as long as they like (about 1 minute per trial, 28 trials in 30 minutes) |
The algorithm, by expected information gain; a trial is five beats plus a response |
Measure |
Circular consistency of the chosen delays across trials (mean resultant length), tested against a Monte Carlo null |
Posterior over the location, width, floor and peak of a response envelope; credible intervals; posterior evidence that the envelope rises at all |
Width of the sensation |
Not separable from consistency |
An explicit parameter, |
Lapses and bias |
Not modelled; confidence weighting offered instead |
Floor and peak parameters absorb lapses, and are marginalized out |
Stopping |
Fixed 28 trials, justified by simulation (15 to 20 enough for classification) |
Credible-interval width on |