Millisecond Forums

Please check the data file of the runs that "[quit] unexpectedly after two human throws" and report what the response column indicates for the player 2 trials (human player).

Hi Supreme Being,

Sorry for the super delayed response. When Cyberball quits unexpectedly, player 2 response = 0 and the values.recipient = 2
However when I rerun the trial and it completes, the response = player3 label and values.recipient = 3
It seems that it quits when player 2 gets the ball for the first time and the subject doesn't have a chance to choose.
There doesn't seem to be a rhyme or reason. I just tried to get it to quit and it quit on the 4th trial. Then I tried it again and then it quit on the 11th trial. Any thoughts?

Below is the actual data of the 4th trial

datetimesubjectblocknumtrialcodetrialnumvalues.playervalues.recipientresponselatencyvalues.count_ignore2
08261623:03:39 1 1 name 2 0 0 b 3736 0
08261623:03:39 1 1 wait 3 0 0 0 3533 0
08261623:03:39 1 2 start 1 1 3 0 2000 1
08261623:03:39 1 2 1to3 2 3 1 0 4000 2
08261623:03:39 1 2 3to1 3 1 2  0 2500 0
08261623:03:39 1 2 1to2 4 2 2 0 2368 0
08261623:03:39 1 3 end 1 2 2 0 3000 0

<trial 1to2> -- the 6th trial in the data excerpt you posted -- is the human player trial. It should not be possible for its response to be 0 (i.e, no response at all), unless that <trial> has a /trialduration or /timeout specified (it doesn't in the original script).

At any rate, if the response to the trial isn't any of those expected via its /validresponse and /branch attributes

<trial 1to2>
...
/ validresponse = (player1, player3, player1label, player3label)
/ontrialend = [if (trial.1to2.response == "player1" || trial.1to2.response == "player1label") {values.recipient = 1; values.count_player2throws += 1; values.count_21 += 1; values.sumrt_21 += trial.1to2.latency; values.count_121 += 1; values.sumrt_121 += trial.1to2.latency}]
/ontrialend = [if (trial.1to2.response == "player3" || trial.1to2.response == "player3label") {values.recipient = 3; values.count_player2throws += 1; values.count_23 += 1; values.sumrt_23 += trial.1to2.latency; values.count_123 += 1; values.sumrt_123 += trial.1to2.latency}]
/ branch = [if (values.recipient == 1) trial.2to1]
/ branch = [if (values.recipient == 3) trial.2to3]
</trial>

i.e., selecting one of the other "computer" players (player 1 and 3) to throw the ball to, the script has nowhere to go and ends. No recipient for the human player's throw is chosen then (cf. values.recipient in the above), i.e. it remains at 2 as is the case in your data:



That is not a condition covered by the /branch logic in the script (because it should not arise in the first place).

I've attached a slightly modified script that performs some additional checks on the response received via /isvalidresponse in the two human player trials (<trial 1to2> and <trial 3to2>. Could you perform a couple of test runs and check if you still see the same issue (no response recorded in human player trials) or any new / different one? Thanks.

Supreme Inquisitive Deity,

It works. Thank you so much. 

Brian

Good morning Brian and Supreme Being,

In a similar train of thought, I am having inconsistency issues with my configuration quitting prematurely and not recording the data. I've had several conversations with one of the programmers (John Downing) about why this may be occurring and he's recommended publishing the configuration online. Unfortunately, attempting to publish online has led every time the pop up of required custom schedules not being assigned properly and every time I attempt to correct this through modifications in the configuration specific to the custom schedules tabs, I end up in the same situation with the pop after clicking the publish button. 

I wonder what advice you may have towards which specific custom schedules either might and/or should be assigned so that I am able to publish the configuration online. According to John, this is a method 90% of experimenters use and the standalone desktop method is very low in use. Furthermore, is there a particular method you may recommend to save the overall configuration after customizing the throw schedule, such that that particular throw schedule remains engaged after X-ing out of the "customize schedules and throws" window of the "customize" tab? Every time I attempt to keep my saved condition engaged in the customize schedules and throws window by saving the overall configuration, it disengages and looks as though it is not saved as the active condition to be engaged in the game. I wonder if this is why the pop-up at the time of selecting the publish configuration under the online/web tab continues to occur, regarding the required custom schedules for each condition not being included or in place properly.

In this regard, is it that all available conditions under the "conditions and random assignment" tab of the configuration app, after selecting the saved condition one has created (schedule, welcome message, connecting message, background image, spectate box, post expt. msg., and post expt. URL) must be included in order to publish the configuration online? Particularly, is that the post expt. URL must be included? If yes to this latter question about post expt. URL, is there a URL that experimenters may collectively publish their configurations and/or results to such that they don't have to create a URL specific to their experiment?

I am hopeful your knowledge towards these questions may guide me in the right direction as I have been struggling a bit through data collection these past couple of weeks. 

Thank you for your time.

Aaron

Hi Aaron,

From reading your description, I'm under the impression that you are referring to an entirely different Cyberball implementation than the one under discussion here. This thread deals with the Cyberball script(s) available for Inquisit https://www.millisecond.com/download/library/cyberball/

At least I am not familiar with the implementation you are talking about, and hence am unable assist with your questions about that implementation.