ILMS Home > Create an Experiment
The description file is part of the Experiment file. The description file specifies what the participant will see and how they can respond. Each row is a trial, each column specifies aspects of the trial.
The fields can be created using the experiment creator, or loaded from a csv (comma-seperated values) file. The csv file can be created in a spreadsheet like Excel (save as a “csv” filetype).
The csv file must not include the “Round” field displayed in the experiment creator.
The csv file’s first column must be the Stimulus field (see below). Other fields can be in any order.
The first row of the csv file must include the names of the fields.
Columns in the csv file must have the same format for all rows.
Here is a list of fields with their function and permitted values. Defaults are set if the values are not specified.
| Field Name | Function | Value |
| Stimulus | The stimulus to display | Number corresponding to number in Mapping file (starts from zero) |
| RoleSwitch | Whether the participants switch speaker/listener roles | 1 = switch roles before trial
0 = keep same roles as last trial |
| NumOptions | The options in the context given to the listener | EITHER a single number which specifies the number of options to show, chosen randomly. OR many numbers separated by a dash (-) which specify e.g.: 1-23-17-3 (actual order of presentation is |
| ContextSet | The set of stimuli from which to draw NumOptions – 1 options for the context. The context will always include the target stimulus | Either blank = this is ignored
OR many numbers separated by a dash (-) which specify e.g.: 1-23-17-3 (actual order of presentation is |
| Context | The exact context of the round. This is usually set by the program when running the experiment | ArrayList of numbers relating to meaning indices. |
| Network | Specifies the network of the participants. This is really for experiments with more than two participants. | Lookup table (Hashtable) mapping a client’s number to their partner’s client number. Defaults to participant 0 and 1 being linked. |
| Training | Is this a training or test trial? | 1 = Training
0 = Test |
| ResetScore | ResetScore to zero before test round | 1 = reset score to zero
0 = don’t reset score |
| ImageWait | During training, the length of time the meaning appears on its own before the signal appears. |
Number of milliseconds |
| WordWait | During training, the length of time the image and signal appear together. |
Number of milliseconds |
| BetweenWait | The length of time between trials (blank screen) | Number of milliseconds |
| SpeakersChoicesType, SpeakersTargetType, ListenersChoicesType |
The format of the options given to the speaker to transmit to the listener |
0 = signals
1 = meanings 2 = typed input 3 = Drawn |
| Feedback | Whether the participants get feedback on their performance | 0 = no feedback
1 = feedback |
| MeaningChoiceFeedback | Whether the participants get to see the option their partner thought was intended in the Double Feedback experiment | 0 = no feedback
1 = feedback |
| FeedbackWait | Length of time feedback is displayed for | Number of milliseconds, defaults to 3000 |
| SpeakerSeesContext | Whether the speaker can see the context that will be presented to the listener |
0 = Speaker can’t see context (default)
1 = |
| PartBreak | An extended break between trials. | 0 = no break
> 0 = number of seconds to give participant a break. When PartBreak >0, a message is displayed, but no |
| Message | Display a message to the participant. | blank =
If PartBreak is zero, don’t If PartBreak is more than zero, Not blank (e.g. “You’ll now do a training round” ) = Display the specified message WARNING: Including commas in this field can corrupt the
When PartBreak >0, a message is displayed, but no |
| ExtraImages | Display an image in a pop-up window. | if partBreak = 0, ignore
else: URL of image = display this image for PartBreak seconds in a seperate frame. Full URL required. |
| ForSending | Add response to list of updates of language to send to another user |
In conjunction with Training=0, i.e. a test round:
0 = don’t use this output to update the langauge 1 = use the speaker’s output to update the language (but 2 = use the most recent signal to update the language (but not immediately) |
| SendSignals | Update signals in mapping with responses indicated by ForSending, and synchronise with other participant. |
0 = don’t send signal (must be set to 0 for training/test)
1 = update and synchronise language with the output from 2 = update and synchronise language with the output from -1 (negative one) = update the langauge with the most recent signals used, regardless of who was the speaker. This row in the specification file will not run a training |
| DynamicStimulus | Indicate whether to use animated stimuli | 0 = don’t use dynamic stimuli (default)
1 = use dynamic stimuli (currently, you still need to Animated stimuli, following Kirby, Cornish & Smith. Implementation in progress. The stimuli number is converted into a 3×3 compositional 0 -> 0,0,0 1 -> 0,0,1 2 -> 0,0,2 3 -> 0,1,0 … 7 -> 0,2,1 |
| FinalTest | Used when creating a new generation with the ExperimentCreator. This has no effect on the experiment, only the output file. | 0 = don’t use this to update the language in the next generation
1 = update the signal for the target meaning with the response from this round Animated stimuli, following Kirby, Cornish & Smith. Implementation in progress. The stimuli number is converted into a 3×3 compositional 0 -> 0,0,0 1 -> 0,0,1 2 -> 0,0,2 3 -> 0,1,0 … 7 -> 0,2,1 |
| AIWait | Time for AI client to wait before responding | Number of milliseconds, defaults to 5000,
A random value up to 10% is added or subtracted to avoid a |