Use guided generation to create any experiment with SigmundAI

A typical visual-working-memory task where circles of different colors need to be remembered. One circle is probed by presenting a circle of either the same or a different color at the original circle's location. The participant responds with a same/different judgment. Set size is varied.

I have a challenging task for you! We're going to implement an OpenSesame experiment from scratch. We'll break the process down in a few steps to make things manageable. In the current step, your task is to expand the description of the experimental task so that it contains all the details that we need to implement it.

Here are some things to consider for the trial structure:

- The way in which participants respond
- The timing of the stimuli
- The appearance of the stimuli
- The layout of the stimuli
- Whether participant feedback is provided and if so how
- Any stimuli that are not explicitly mentioned but that should be included
- Blank or fixation displays in between stimuli (if there is a fixation dot, it is usually shown throughout the trial)

Here are some things to consider for the experiment structure:

- Independent and dependent variables
- Whether or not there is practice phase
- Whether or not the trials are subdivided into blocks of trials
- The number of trials and (if applicable) blocks
- Any welcome, instruction, and goodbye screens

Please also include any additional information that you think is relevant.

<experimental_task_description>
A typical visual-working-memory task where circles of different colors need to be remembered. One circle is probed by presenting a circle of either the same or a different color at the original circle's location. The participant responds with a same/different judgment. Set size is varied.
</experimental_task_description>

Please reply with a comprehensive description. Do not save it as a note yet, because we are first going to review it.

Awesome! Please save the comprehensive description as a persistent note so you don't forget. Do not summarize it, but save it in full.

Moving on! Here's your next task:

- Draft a JSON specification based on the narrative description of the experiment that you just saved as a note.
- Each dict in the JSON should correspond to a single OpenSesame item. You can use the following items: [loop, sequence, sketchpad, feedback, synth, sampler, keyboard_response, mouse_response, logger, inline_script, inline_javascript, form_multiple_choice, form_text_display, form_text_input, reset_feedback]
- Sketchpad items are prepared in advance, and can therefore not take into account variables that are defined after their prepare phase. Therefore, to show displays with variable content, you often want to use feedback items instead, but only for displays that are not time-critical.
- Do not use multiple unlinked logger items, because this will result in messy log files. Instead, use linked copies of the same logger.
- Do not put reset_feedback items in the main experiment sequence, because this is not compatible with OSWeb.
- Don't include any implementation details for the items. We'll get to that later. For now, just provide a description, name, and type. For items that are part of a sequence, you can also provide a run-if expression. For items that are linked copies of items that occur elsewhere, provide details for the first occurrence only, and provide only the name for subsequent occurrences (items with the same name are linked copies).
- Items that have child items have an additional 'items' field. A loop always has a single sequence as a child item. A sequence generally has multiple child items of various types.
- Please use linked copies of items whenever possible. To indicate that an item is a linked copy of another item, simply re-use the same name and add a `linked` field that is set to `true`.
- The example JSON specification below shows the general idea, but is highly simplified. Your JSON specification will likely be much more elaborate.

<json_example>
{
  "name": "experiment",
  "type": "sequence",
  "description": "A one line description of the task",
  "items": [
    {
      "name": "task_description",
      "type": "notepad",
      "description": "The full narrative description of the task goes here."
    },
    {
      "name": "welcome",
      "type": "sketchpad",
      "description": "Brief welcome message"
    },
    {
      "name": "block_loop",
      "type": "loop",
      "description": "A block of trials. Independent variables are defined here as well.",
      "items": [
        {
          "name": "trial_sequence",
          "type": "sequence",
          "description": "Sequence for a single trial",
          "items": [
            {
              "name": "fixation",
              "type": "sketchpad",
              "description": "Shows black fixation cross (+) for 500 ms"
            },
            {
              "name": "target_display",
              "type": "sketchpad",
              "description": "Shows target stimulus while the fixation cross remains visible"
            },
            {
              "name": "keyboard_response",
              "type": "keyboard_response",
              "description": "Collects participant response to target stimulus"
            },
            {
              "name": "correct_feedback",
              "type": "sketchpad",
              "description": "Shows green fixation dot for 500 ms after correct response",
              "run_if": "correct == 1"
            },
            {
              "name": "fixation",
              "linked": true
            },
            {
              "name": "logger",
              "type": "logger",
              "description": "Logs all trial data"
            }
          ]
        }
      ]
    }
  ]
}
</json_example>

Please reply with the JSON specification for the narrative description of the experiment. Do not save the JSON specification as a note yet, because we are first going to review it.

Beautiful, well done Sigmund! Please save this as another persistent note so you don't forget. Do not summarize it, but save it in full.

We're now ready to implement the experiment. A few pointers:

- Save these instructions as a persistent note so you don't forget.
- Do *not* inspect items of the current experiment. They're not relevant, because we're going to completely overwrite the current experiment.
- Do *not* inspect the general script of the current experiment for the same reason.
- Before writing the experiment script, call `opensesame_get_syntax_documentation` with `save_as="note"` to get all relevant documentation.
- Finally, write the new experiment as a single complete general script and pass it to `opensesame_update_general_script`.

This is a challenging task, but I know you can do it. Let's go!

A typical visual-working-memory task where circles of different colors need to be remembered. One circle is probed by presenting a circle of either the same or a different color at the original circle's location. The participant responds with a same/different judgment. Set size is varied.

A Posner cuing paradigm with a central cue and a letter-discrimination task.

An AX continuous performance task.