Heartbeat Counting task

The Heartbeat Counting task (HBC) [Schandry, 1981, Dale and Anderson, 1978] compares the number of heartbeats a participant reports during a fixed interval with the number recorded. It is an established task with a large literature, in which the score is commonly interpreted as interoceptive accuracy.

A good score can also result from a plausible estimate of heart rate, without reliable perception of individual beats. What heartbeat counting can establish reviews the evidence and its implications for interpreting a counting score.

What a trial looks like

A tone marks the start of the window. The participant silently counts the heartbeats they feel while the pulse is recorded. A second tone marks the end. They then type the number they counted and press enter, and rate their confidence from 1 (“Guess”) to 7 (“Certain”) on the keyboard.

One HBC trial. The window lengths and the 1 to 7 confidence scale are the defaults.

The default version, taskVersion="Garfinkel", presents six counting windows of 25, 30, 35, 40, 45, and 50 s in random order. The taskVersion="Schandry" version presents three windows of 25, 35, and 45 s in that order, following a minute of rest and separated by 30 s rest periods.

With runTutorial=True, the session begins with instruction screens. Every version also records the pulse during a 30 s rest before the first window. The default version takes about 4 minutes.

What you need

• A pulse oximeter on a USB port (the Nonin 3012LP Xpod by default), or an ECG or PPG device streaming over Lab Streaming Layer. Recording devices covers both and how to check the oximeter before a session.

• Speakers or headphones for the start and stop tones.

• A keyboard.

Running it

Copy wrappers/hbc.py from the repository next to your data folder and run it:

python hbc.py

A dialog asks for the participant and session identifiers, the serial port the oximeter is on (python -m cardioception.check_device --list lists the ports), the setup and the language. The same in a script:

from cardioception.HBC.parameters import getParameters
from cardioception.HBC.task import run

parameters = getParameters(
    participant="Subject_01", session="HBC", serialPort="COM3",
    setup="behavioral", taskVersion="Garfinkel")
run(parameters, runTutorial=True)
parameters["win"].close()

With an LSL device, pass recorder=LSLRecorder(name="PLUX", channel="ECG2", signal_type="ecg") instead of a port, as on the HRD page. setup="test" replays a recorded pulse, for checking a script without a device. Escape ends the session.

The main settings

Argument	Default	Meaning
taskVersion	"Garfinkel"	"Garfinkel", "Schandry", or "test" (a 5 s rest and a 5 s count)
language	"english"	"english" or "german"; cardioception.HBC.languages.available() lists them
confidenceScale	DISCRETE_1_7	A cardioception.scales.ConfidenceScale. The end labels follow the language
setup	"behavioral"	"behavioral" reads the oximeter; "test" replays a recorded pulse
serialPort	"COM3"	The port the oximeter is on
resultPath	data	Root of the run directories
overwrite	False	Allow writing into a run directory that already holds results
seed	drawn	Seeds the order of the windows; recorded in the manifest
triggers	none	{event: callable} for trialStart, listeningStart, listeningStop, confidenceStart, confidenceStop and trialStop

To define a custom schedule, edit parameters["times"] and parameters["conditions"] after getParameters returns. These contain the window lengths and the corresponding conditions ("Rest" or "Count").

What it saves

A session writes to data/sub-<participant>/ses-<session>/run-<YYYYMMDD-HHMMSS>/, under resultPath when it is given.

File	Contents
*_final.txt	One row per window
*_ppg-rest.txt, *_ppg-<n>.txt	The pulse recorded during the rest and during window n, with the trigger codes in Channel_0 (1 at the start tone, 2 at the stop tone)
manifest.json	The settings, the seed and the version, written before the first trial
*_behaviour.txt, session.log	The rows written after every window, and the session’s log

The columns of *_final.txt:

Column	Meaning
nTrial	The window number, from 0
Condition	Rest or Count
Duration	The length of the window in seconds
Reported	The number the participant typed
Confidence, ConfidenceRT	The rating on the scale shown, and its response time. ConfidenceUnit holds the rating on 0 to 1, and the scale columns describe the scale

The task saves the pulse recording without counting the beats. Obtain the recorded count for each window by detecting beats in its pulse file.

Scoring

Detect the beats in each *_ppg-<n>.txt file, count them, and score each window with the accuracy of Hart et al. [2013]:

\[ Accuracy = 1-\frac{\left | N_{real} - N_{reported} \right |}{\frac{N_{real} + N_{reported}}{2}}\]

A score near 1 is consistent with feeling every beat, and also with knowing one’s resting heart rate; the score cannot separate the two.

Where to go next

• What heartbeat counting can establish: the evidence on what a counting score identifies.

• API: every argument of getParameters.