I was expecting to see an error message when I tried to use a semi-colon delimited spreadsheet, but it appears that alternative delimiters do still work. In addition to semi-colon and tab, pipe symbols (|) also work online but not locally.
Cells containing delimiters (such as commas) should be enclosed in quotes.
This is one of the most common errors for experiments which run locally but fail online. In the case of a spreadsheet the name of the spreadsheet is mentioned twice. In the case of an image, you see the name of the image component (ImageStim) and the name of the file (resources).
At the start of an online experiment you usually see a download bar for the resources.
However, at this stage PsychoPy only downloads files it can identify, either because they are specified in components, loops or spreadsheets. Resources it can’t identify automatically need to be added to the list manually. The easiest way to do this is via Experiment Settings / Online / Additional Resources.
Add resources using the plus and remove them with the minus. The three dots offer to add a custom item, but when I last explored this option it didn’t seem to work.
If you don’t want to download all resources, for example because each participant sees a subset of the stimuli, then you can download addition resources in code.
A single-select matrix in Pavlovia surveys can easily be set to require a single response or all responses, but custom response pattern requirements are also possible. In all cases the required toggle at the bottom of the question should be switched on.
Tick Validation / Require answer for all rows
This is the default. Validation / Require answers for all rows is unticked but in this case I have added a custom “Required” error message
For this option I create a separate read-only expression question with the expression
{block_1/question3.1} notempty + {block_1/question3.2} notempty + {block_1/question3.3} notempty + {block_1/question3.4} notempty +
{block_1/question3.5} notempty + {block_1/question3.6} notempty + {block_1/question3.7} notempty + {block_1/question3.8} notempty +
{block_1/question3.9} notempty + {block_1/question3.10} notempty
In this demo the expression is visible. If you hide it remember to change “clear the value if the question becomes hidden” to None.
Then add a custom validation expression and error message such as {block_1/question4} > 4
This technique is most useful if the final entry of the single-select matrix is an Other option, which you want to be optional. In this case you could exclude that row from the expression question.
This is my suggested route to make your own copy of an existing project, whether it is public or has been shared with you.
Experiments already shared with you will appear in your Experiments tab on Pavlovia. Alternatively, search for public projects in the Explore tab.
Click on View code to go to the GitLab page.
This can either be your main account or a group. If you are forking your own project then you will have to select a group (unless you are forking from a group) since you can’t have two projects with the same name in the same namespace, and forking doesn’t let you edit the project name.
In this example I forked https://pavlovia.org/demos/face_api to https://gitlab.pavlovia.org/Wake/delay_discount so I next edit the URL to https://pavlovia.org/Wake/delay_discount.
Set the status to Piloting (or Running) and turn off “save incomplete results”. If you can see a platform version you can pilot it.
Log in to Pavlovia via Builder, if you haven’t already and select Search projects
You can now start making edits and sync them to update your online version. Enjoy!
If you use Besample to recruit participants then you have to display a completion code on screen. To avoid showing this before the participant has actually completed the study, you can now redirect to my latest VESPR tool
https://moryscarter.com/vespr/debrief.php
If you wish to display a completion code, please call this page using the URL https://moryscarter.com/vespr/debrief.php?participant=10&session=208&id=besample
For more details about daisy chaining, see Wakefield's Daily Tips - #35 by wakecarter
As of 28th March I’ve added email confirmation and custom logos to the debrief page.
There are two basic methods of playing media (sounds and videos) in a Browser.
The larger the file, the better option 2 is compared with option 1.
In both cases the media file needs to be hosted somewhere.
For option 1, the easiest place to host your media files in on Gitlab. For example, I have a PsychoPy experiment called Stressful Stroop which plays a buzzer sound if you make a mistake. The run link for the experiment is https://run.pavlovia.org/Wake/stressful-stroop/ and the buzzer sound is BUZZER.mp3 so the URL for the buzzer is https://run.pavlovia.org/Wake/stressful-stroop/BUZZER.mp3.
I can add this to a Pavlovia Survey using the following code for mp3 and wav files:
<audio controls>
<source src="https://run.pavlovia.org/Wake/stressful-stroop/BUZZER.mp3" type="audio/mp3">
</audio>
Parameters for the audio tag are autoplay, controls, loop, muted, preload and src. Autoplay will not work on the first page.
Similarly, for videos you can use the video tag for mp4 and webm files.
For larger file, I would recommend hosting on Google Drive so they can be streamed in an iframe. Here are my slides on how to do this in Qualtrics, which will also work in Pavlovia Surveys.
Share outside your domain.
Copy link
Identify the unique Google Drive code for the media.
That code appears as xxx in the following html.
<iframe src="https://drive.google.com/file/d/xxx/preview" width="640" height="480"></iframe>
Edit the width and height as appropriate.
You can also try with iframe embed codes from other platforms. For Vimeo, videos can be set to autoplay and the code is
<iframe src="https://player.vimeo.com/video/486333942?autoplay=1" width="640" height="360" frameborder="0" allow="autoplay; fullscreen" allowfullscreen></iframe>
As a final note, one way of having an audio file is to make the size of a video too small to see.
I used to use negative width to present a mirror image of a visual stimulus. Unfortunately, this stops the stimulus being clickable with a mouse online.
Instead, use flipHoriz in a code component. For example,
if trials.thisN%2:
image.flipHoriz = True
else:
image.flipHoriz = False
The polygon fix is to have an invisible polygon to click on instead.
Do not use the Dynamic URL in the Survey Complete section of the survey properties. This is not compatible with Pavlovia Surveys
Instead you should use the Experiment Flow section of the overview page.
Here I have set the Completion URL to
"https://" + {block_1/question4} + {block_1/dissent}
dissent is an expression based on the consent question
iif({question1} = false,"google.com","")
question4 is an expression based on a later question
iif({question2} = 1,"moryscarter.com/vespr",iif({question2} = 2,"pavlovia.org",""))
Note that in both cases I have set the value to be an empty string “” if the condition is not met.
N.B. The survey completion URL must be blank if your survey is embedded.
If you create a survey then you edit the JSON code directly. This is not generally advisable, unless you want to copy and paste a question or page from one survey (or survey block) to another.
Here is the JSON code for a block of a survey containing three questions spread across two pages.
{
"title": "JSON example",
"pages": [
{
"name": "page1",
"elements": [
{
"type": "radiogroup",
"name": "question1",
"choices": [
"Item 1",
"Item 2",
"Item 3"
]
},
{
"type": "matrix",
"name": "question2",
"columns": [
"Column 1",
"Column 2",
"Column 3"
],
"rows": [
"Row 1",
"Row 2"
]
}
]
},
{
"name": "page2",
"elements": [
{
"type": "boolean",
"name": "question3"
}
]
}
]
}
The survey is defined as a nested dictionary.
At the survey level we can see two keys: “title” and “pages”.
“pages” contains a list of two dictionaries, one for each page.
Each page has keys “name” and “elements”.
“elements” contains a list of dictionaries, one for each question.
Each question has keys “type” and “name” plus optional additional keys depending on the question type: “choices”, “columns” and “rows”.
If you want to move a section of your survey from one location to another, you need to identify the relevant page or question dictionary.
If a question has display logic then the question dictionary also have a “visibleif” key, such as.
"visibleIf": "{question1} = 'Item 1'",
Pages can also have a “visibleif” key.
If you edit the JSON code and create invalid dictionaries, for example by having duplicate question names, then you will not be able to move off the JSON view until you have fixed the issue.
N.B. The block names are not shown in the question names in the JSON files but are needed in logic referring to more than one question.
Today I wrote (with the help of chatGPT) a Python script to extract all of the expInfo variables in Builder files across a set of folders. In the script below if looks in every folder in my Documents/Pavlovia folder and creates an Excel file containing the Key (expInfo variable name), Type (STRING, NUMBER or BOOLEAN), default value, List (if the value is a list of options), Folder (for if your group your experiments) and Experiment (based on the folder name, not the experiment name).
import pathlib
import os
import re
import fnmatch
import pandas as pd
import ast
# Modify this line to locate your experiment files
experiments_folder = pathlib.Path.home() / "Documents" / "Pavlovia"
# Regex to find the Experiment info dictionary
pattern = r'<Param val="([^"]+)" valType="code" updates="None" name="Experiment info"/>'
data = [] # List to store extracted rows
def is_number(value):
"""Check if a value is a number, even if stored as a string."""
try:
float(value)
return True
except (ValueError, TypeError):
return False
for path, dirs, files in os.walk(experiments_folder):
path = pathlib.Path(path) # Convert to Path object
if len(path.parts) > len(experiments_folder.parts): # Ignore chosen folder
#for filename in fnmatch.filter(files, "*.psyexp"): # Match all .psyexp files
for filename in [f for f in fnmatch.filter(files, "*.psyexp") if not fnmatch.fnmatch(f, "*_legacy.psyexp")]:
filepath = path / filename
experiment_name = path.name # Folder name as experiment identifier
with open(filepath, "r", encoding="utf-8") as file:
try:
content = file.read()
except (SyntaxError, ValueError) as e:
print(f"❌ Error opening {filepath}: {e}")
matches = re.findall(pattern, content)
if matches:
print(experiment_name)
param_value = matches[0] # Assume one match per file
clean_param_value = param_value.replace(""", "\"")
try:
param_dict = ast.literal_eval(clean_param_value)
for key, value in param_dict.items():
required = "|req" in key
config = "|cfg" in key
clean_key = key.replace("|req", "").replace("|cfg", "") # Remove markers
# Default Options column as empty string
options = ""
# Detect lists inside strings
if isinstance(value, str) and ((value.startswith("[") and value.endswith("]")) or "," in str(value)):
try:
value = ast.literal_eval(value) # Convert string to list if possible
except (SyntaxError, ValueError):
pass # Keep as a string if conversion fails
# Determine Type and adjust Value column
if isinstance(value, (list, tuple)):
options = str(list(value))
first_item = value[0] if value else None # First item or None
value_type = "NUMBER" if is_number(first_item) else "STRING"
value = first_item # Store first item in Value column
elif isinstance(value, list):
options = str(value)
first_item = value[0] if value else None # First item or None
value_type = "NUMBER" if is_number(first_item) else "STRING"
value = first_item # Store first item in Value column
elif is_number(value): # Detect numbers stored as strings
value_type = "NUMBER"
value = float(value) if "." in str(value) else int(value)
elif value == "True":
value_type = "BOOLEAN"
else:
value_type = "STRING"
folder_name = path.parent.name
data.append((clean_key, value_type, required, config, value, options, folder_name, experiment_name))
except (SyntaxError, ValueError) as e:
print(f"❌ Error parsing dictionary in {filepath}: {e}")
# Convert to DataFrame
df = pd.DataFrame(data, columns=["Key", "Type", "Required", "Config", "Value", "List", "Folder", "Experiment"])
# Replace None with empty string explicitly
df["List"] = df["List"].fillna("")
# Aggregate assessments into lists
df_grouped = df.groupby(["Key", "Type", "Required", "Config", "Value", "List", "Folder"], dropna=False)["Experiment"].apply(list).reset_index()
# Save to Excel
output_path = experiments_folder / "experiments_summary.xlsx"
df_grouped.to_excel(output_path, index=False)
print(f"✅ Data saved to {output_path}")
This probably doesn’t have much use if you only have a few experiments but I have quite a lot (as you can probably imagine) and today I wanted to check the ones created for a particular client to see if I had been using the same variable names and defaults across the different experiments.
Required is identified by a |req tag at the end of the key.
Config is identified by a |cfg tag at the end of the key, e.g. offWhite|cfg.
expInfo_variables.py (4.5 KB)
This error seems to occur when you are trying to find something out about an undefined list.
The first step is to identify which list is undefined, using Developer Tools. Next, try to work out why it might be undefined at that point in your code. The answer is usually either a blank cell or row in a spreadsheet or the object is being checked too soon.
For example, this code will fail online either in Each Frame or in End Routine if the routine can end without a key press.
numKeys = len(key_resp.keys)
The fix is to confirm that key_resp.keys exists before checking its length:
if key_resp.keys:
numKeys = len(key_resp.keys)
This error appears when you try to use code which does not have an PsychoJS equivalent. You need to try a different approach.
For example, if you have created a countdown timer, you just need to define the clock in a Both code component (probably the Begin Experiment tab) and edit the JavaScript side.
myClock = core.CountdownTimer(60) # Python
myClock = new util.CountdownTimer(60); // JavaScript
Once defined you can use myClock.reset() and myClock.getTime() as normal.
On the other hand, if you have used visual.Circle in a code component and get TypeError: visual.Circle is not a constructor use visual.ShapeStim instead.
visual.ShapeStim(
win=win,
fillColor='white',
lineColor='white',
vertices=[[0.000,0.866], [0.091,0.861], [0.180,0.847], [0.268,0.824], [0.352,0.791], [0.433,0.750], [0.509,0.701], [0.579,0.644], [0.644,0.579], [0.701,0.509], [0.750,0.433], [0.791,0.352], [0.824,0.268], [0.847,0.180], [0.861,0.091], [0.866,0.000], [0.861,-0.091], [0.847,-0.180], [0.824,-0.268], [0.791,-0.352], [0.750,-0.433], [0.701,-0.509], [0.644,-0.579], [0.579,-0.644], [0.509,-0.701], [0.433,-0.750], [0.352,-0.791], [0.268,-0.824], [0.180,-0.847], [0.091,-0.861], [0.000,-0.866], [-0.091,-0.861], [-0.180,-0.847], [-0.268,-0.824], [-0.352,-0.791], [-0.433,-0.750], [-0.509,-0.701], [-0.579,-0.644], [-0.644,-0.579], [-0.701,-0.509], [-0.750,-0.433], [-0.791,-0.352], [-0.824,-0.268], [-0.847,-0.180], [-0.861,-0.091], [-0.866,0.000], [-0.861,0.091], [-0.847,0.180], [-0.824,0.268], [-0.791,0.352], [-0.750,0.433], [-0.701,0.509], [-0.644,0.579], [-0.579,0.644], [-0.509,0.701], [-0.433,0.750], [-0.352,0.791], [-0.268,0.824], [-0.180,0.847], [-0.091,0.861]],
pos = [0,0],
depth = 50,
size=.3,
opacity = 1
)
This error is usually caused by having a blank cell in a spreadsheet which your experiment tries to use as a component parameter that needs to be a number, such as the position of a visual stimulus.
If you can’t find a blank cell, it can be more difficult to trace the error because the line itself doesn’t appear in the Developer Tools.
