Hi Thanks for your reply. Regarding the order of trials /ITI, when using a uniform distribution we just randomise both the trial (i.e. condition) order and the ITI order separately. If you are using a negative exponential do you need to do some pseudo-randomisation to ensure that the longer ITI are distributed between the conditions equally? I notice your Excel tool is set to only have 1 or 2 of the longest ITI within each block – if you have 8 conditions this could lead to many conditions never having a trial with the longest ITI if the allocation of the ITI to trials is just randomised. Thanks Rob From: MCLAREN, Donald [mailto:[log in to unmask]] Sent: 14 June 2015 21:29 To: Robert Hoskin Cc: SPM Subject: Re: Optimal timing of stimuli during event related designs You can probably use a short interval between the cue and outcome. You have 4 cues and 8 outcomes, so each cue will be followed by 1 of 2 outcomes. This will help you separate the cue from the outcome. The other issue is whether the response to cue stops when the cue stops or continues until the outcome screen comes on. Yes. A negative exponential can help with the determining the timing gaps. One you have the list of all possible ITIs, you'll need to decide what order your trials occur and what order your ITIs should occur. I've attached a very old program I used to use before optseq2 was released. It'll give you the frequencies of each ITI. The rest is up to you (stimulus order and jitter order. Best Regards, Donald McLaren ================= D.G. McLaren, Ph.D. Research Fellow, Department of Neurology, Massachusetts General Hospital and Harvard Medical School Postdoctoral Research Fellow, GRECC, Bedford VA Website: http://www.martinos.org/~mclaren Office: (773) 406-2464 ===================== This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is intended only for the use of the individual or entity named above. If the reader of the e-mail is not the intended recipient or the employee or agent responsible for delivering it to the intended recipient, you are hereby notified that you are in possession of confidential and privileged information. Any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited and may be unlawful. If you have received this e-mail unintentionally, please immediately notify the sender via telephone at (773) 406-2464 or email. On Thu, Jun 11, 2015 at 1:58 AM, Robert Hoskin <[log in to unmask]<mailto:[log in to unmask]>> wrote: Hi, thanks for your replies. The trial structure is Fixation (500ms) Cue (1000ms) <Interval> Outcome (750ms) < ITI> The cue give the participant the probabilities of various outcomes. There are 4 cues, 2 of which are associated with outcomes A and B (probabilities 75/25% and 25/75% respectively), and the other 2 are associated with outcomes A and C (75/25 and 25/75 again). The relationship between the cue and outcome is not therefore orthogonal so I think we would need to be able to separate the responses to the cue from the responses to the outcome. I think, as Helmut says, we would want the interval between cue and outcome to be shorter than the ITI to maintain the integrity of the trial. Given scanning time we could really do with an average trial length of 10s max, which is difficult to achieve with two intervals. Perhaps 2 - 4 for the interval and 3 - 7 for the outcome? Or would that be too short to distinguish the responses to cue and outcome. I've seen papers where the ITI is jittered using a non-uniform distribution (e.g. the number of trials with each of the ITI lengths is different). For example Ziaei et al (2014) 'Brain systems underlying attentional control...' use an ITI distribution of: 42% 1.5 s, 28% 3 s, 14% 4.5 s, 12% 6 s, and 4% 7.5 s). There is an earlier paper (Hagberg et al (2001) 'Improved detection of event-related....') which seems to suggest using simulated data that Geometric or exponential distributions of ITI, biased towards shorter ITI (As per Ziaei et al) are better than uniform distributions at Detection and Parameter estimation, as long as the range of ITIs in the distribution is quite wide (i.e. a small number of long ITIs of say 15s). I can't find any more recent work on this. Is this a common method? Are these distributions what optseq produces? - I couldn't get it to work on my windows machine. Thanks Rob