From 659ce4066e507e531fb459d112e7ac189f262dba Mon Sep 17 00:00:00 2001 From: Nir Moneta Date: Fri, 27 Mar 2020 12:23:56 +0100 Subject: [PATCH 1/2] Update proposal.md --- docs/proposal.md | 54 ++++++++++++++++++++++++++++-------------------- 1 file changed, 32 insertions(+), 22 deletions(-) diff --git a/docs/proposal.md b/docs/proposal.md index 1e09f15..c34fdf9 100755 --- a/docs/proposal.md +++ b/docs/proposal.md @@ -1,31 +1,38 @@ - ### Motivation -In a series of laboratory studies investigating the role of trait anxiety in aversive learning, we found that contingency changes were better tracked by a high trait anxiety group. Behavioral and physiological (SCR) data revealed that high trait anxious individuals had a hightened awareness of the current hidden state of the environment, especially in the later stages of the experiment (Fig 1). This was indexed by more distinct expectancy ratings and anticipatory SCR responses between periods of objectively low and high threat. Furthermore, we found that following changes in the underlying shock contingency, high trait anxious individuals were faster at adjusting their expectation to the new reinforcement level, indexed by switch steepness. Computational modeling revealed that there are generally two groups of participants: those who update their expectations gradually in a trial-by-trial manner, and those that learn that there are two contingency states and switch between them in an abrupt manner. Trait anxiety was associated with a tendency to employ the state switching, rather than the gradual learning strategy. +In a series of laboratory studies investigating the role of trait anxiety in aversive learning we found that contingency changes were better tracked by high trait anxiety group. Behavioral and physiological (Skin Conductance Response, henceforth: SCR) data revealed that high trait anxiety individuals had higher awareness of the current hidden state of the environment, especially in the later stages of the experiment (Fig 1). This was indexed by more distinct expectancy ratings and anticipatory SCR responses between periods of objectively low and high threat. Furthermore, we found that following changes in underlying shock contingnecy, high trait anxious individuals were faster at adjusting their expectation to the new reinforcement level, indexed by switch steepness. Computational modeling revealed that there are generally two groups of participants: those who update their expectations gradually in a trial-by-trial manner, and those that learn that there are two contingency states and switch between them in abrupt manner. Trait anxiety was associated with tendency to employ the state switching, rather than gradual learning strategy. -*Fig. 1: Data show trial-by-trial adjustment of shock expectancy after a switch from high to low (blue) and low to high (red) probability of shock. Lighter colors show the high trait anxious group.* +*Fig. 1: Data show trial-by-trial adjustment of shock expectancy after a switch from high to low (blue) and low to high (red) probability of shock. Lighter colors show high trait anxious group.* ![behavioural_data](behavioural_data.png) -The above findings may potentially explain why we see higher rates of fear relapse in highly anxious individuals. Our group is currently conducting a study investigating whether state switching and high trait anxiety also lead to higher rates of relapse on a session one week later. +(Nir: i would take the legend and put it under the two subplots, otherwise mainly because there is no differnece in the reds on the left, its a bit confusing) -The ongoing global crisis related to the COVID-19 pandemic resembles our experimental design in number of ways: There is an initial fear period which will be followed by a phase of fear extinction and most likely (see below) by a period of fear reinstatement. We therefore see the situation as a unique opportunity to investigate fear learning in a natural setting. +The above findings may potentially explain why we see higher rates of fear relapse in highly anxious individuals. Our group is currently conducting a study investigating whether state switching and high trait anxiety also lead to higher rates of relapse on a session one week later. -### Hypotheses +The ongoing global crisis related to the COVID-19 pandemic resembles our experimental design in a number of ways: There is an initial fear period which will be followed by a phase of fear extinction and most likely (see below) by a period of fear reinstatement. We therefore see the situation as a unique opportunity to investigate fear learning in a natural setting. +(Nir: i would seperate here between what is clear and what is prediction to future re the corona virus) -*Fig. 2: Simulated time-course of threat probability estimates for high and low trait anxiety groups. Blue rectangles highlight periods where theoretical predictions for state-switchers and gradual learners diverge.* -![fear](fear.png) +Nir: im missing here your hypothesis in clear words, something like: +based on the predicted number of corona cases (X et al) we expect a general corresponding increase in anxiety charecterized by a temporal delay (Nir: bad phrasing but you get the point) +Based on our previous finidngs, we thus expect that high anxoues would show even more..... compared to low anxouse.... -In Fig. 2 we show the main predictions based on our lab results. The predictions concern specifically threat-related probability judgments, not subjective fear. +### Hypotheses #### Core -1. High TA participants will recognize a period of safety faster and adjust their *objective* aversive event expectations sooner. -2. High TA will be associated with faster threat probability (based on our lab results) and fear (based on literature) reinstatement once signs of the second wave of the pandemic occur. +1. High trait anxiety participants will recognize a period of safety faster and adjust their *objective* aversive event expectations sooner. +2. High trait anxiety will be associated with faster threat probability (based on our lab results) and fear (based on literature) reinstatement once signs of the second wave of the pandemic occur. #### Exploratory 3. There will be a dissociation between affective and objective assessment of the situation that will be particularly pronounced in highly anxious individuals. +*Fig. 2: Simulated time-course of threat probability estimates for high and low trait anxiety groups. Blue rectangles highlight periods where theoretical predictions for state-switchers and gradual learners diverge.* +![fear](fear.png) + +In Fig. 2 we show the main predictions based on our lab results. The predictions concern specifically threat-related probability judgments, not subjective fear. + + ### Design -We propose a longitudinal study lasting 12 months in which an initial sample of 1000 native English speakers in the UK and the US will complete 12 on-line sessions. In the sessions we will collect questionnaire responses assessing personality traits (trait anxiety, catastrophizing, etc.) current level of anxiety, stress, covid-related fears and event probability estimates. An additional smaller sample of ~100 people, which will be recruited in person, will be added to minimize the risk of complete dropout. +We propose a longitudinal study lasting 12 months in which an initial sample of 1000 native English speakers in the UK and the US will complete 12 on-line sessions during which we will collect questionnaire responses assessing personality traits (trait anxiety, catastrophizing, etc.) current level of anxiety, stress, covid-related fears and event probability estimates. Additional smaller sample of ~100 people which will recruited in person will be added to minimize risk of complete dropout. #### Materials ##### Demographic measures (DMs) @@ -37,7 +44,7 @@ We propose a longitudinal study lasting 12 months in which an initial sample of 1. STAI-TRAIT and STICSA (STAI-TRAIT alternative) 2. BDI-II 3. General Catastrophizing (new questionnaire by Pike et al.) -4. Risk perception/attitute +4. Risk perception/attituted ##### Current state measures (CSMs) 1. STAI-STATE @@ -49,7 +56,7 @@ We propose a longitudinal study lasting 12 months in which an initial sample of #### Session types **1. First Session** - - DMs, PMs and CSMs + - DMs, PMs and CSMs (Nir: i would keep this full description or even just make a plot of a table to show the timeline?) - 25 min **2. Full session** @@ -61,23 +68,26 @@ We propose a longitudinal study lasting 12 months in which an initial sample of - 10 min #### Timing -There is a plan to conduct 12 sessions in total, 4 of which will be first/full sessions while the other 8 will be the shorter check-in sessions. The temporal sampling will be uneven, reflecting the expected progress of the global pandemic. This is to ensure that the critical periods (e.g. resurgence in number of cases expected in Nov/Dec 2020) are sampled with higher frequency. Crucially, the increased sampling frequency will start as soon as the news start reporting a COVID-19 comeback. +There is a plan to conduct 12 sessions in total, 4 of which will be first/full sessions while the other 8 will be the shorter check-in sessions. The temporal sampling will be uneven, reflecting the expected progress of the global pandemic. This is to ensure that the critical periods (e.g. resurgence in number of cases expected in Nov/Dec 2020) are sampled with higher frequency. Crucially, the increased sampling frequency will start as soon as the news start reporting COVID19 comeback. *Fig. 3: Predicted health care demands, UK, next 12 months (based on Anderson et al. 2020, The Lancet)* ![Sampling](sampling.png) #### Payment structure -Participants will be paid at a rate of $9/hr (estimate from Prolific, mTurk can be cheaper). On sessions 3, 6 and 9 they will be paid a $2 interim bonus and on the last session an $8 overall bonus. A participant that completes all 12 sessions will earn $32. If every participant completed the study, the cost would be $32 000. However, an attrition rate is expected. Approximately 20% of the participants are expected to complete the study (attrition shown on figure below), so the estimated cost for the on-line sample is $13 500. Adding 100 in-person recruited participants extends the cost by further $3200, bringing the total to $16 700. +Participants will be paid at rate $9/hr (estimate from Prolific, mTurk can be cheaper). On sessions 3, 6 and 9 they will be paid $2 interim bonus and on the last session $8 overall bonus. A participant that completes all 12 sessions will earn $32. If every participant completed the study, the cost would be $32 000. However, an attrition rate is expected. Approximately 20% of participants are expected to complete the study (attrition shown on figure below), so the estimated cost for on-line sample is $13 500. Adding 100 in-person recruited participants extends the cost by further $3200, bringing the total to $16 700. *Fig. 4: Payment structure of the experiment* ![Payment](payment.png) -As an alternative, in order to motivate people more to complete the sessions, we can raise the interim bonus to $5 and the final bonus to $10. This is up to discussion. The cost in that case would be $23 000 (but nearly $50k if they completed all 12 sessions). +As alternative, to motivate people a bit more to complete, we can raise interim bonus to $5 and the final bonus to $10. This is something to discuss. The cost in that case would be $23 000. (but nearly $50k if they all completed the the 12 sessions). #### Questions This section contains all questionnaires newly designed for this study. +(Nir: make clear what will the participants will see and what is explanations for the researcher/ethics etc..) + + --- ##### 1. Factual COVID-related information [Factual COVID] @@ -90,7 +100,7 @@ A. *Please tick what applies to you in relation to the coronavirus:* - [ ] A person close to me fell seriously ill or died. -B. *Please indicate to what degree the following applies to you __as a consequence__ of the coronavirus:* +B. *Please indicate to what degree do the following apply to you __as a consequence__ of the coronavirus:* [Responses: Does not apply 0:10 Strongly applies] @@ -123,11 +133,11 @@ A. *Due to the coronavirus I am __currently__:* 6. *worried that a close person will get infected.* 7. *worried that a close person will suffer serious medical issues or die.* -B. *Right now, we are in a period of relative danger.* +B. *Right now, we are in a period of relative danger.* C. *I was surprised when the coronavirus became pandemic in my country.* D. *When the pandemic broke out, I was very scared.* E. *Many people are over-reacting.* -F. *The virus is not as dangerous as is often portrayed.* +F. *The virus is not as dangerous as it is often portrayed.* G. *The virus was made in a lab.* H. *Right now, we are in a period of relative safety.* @@ -144,11 +154,11 @@ A. *Please try to objectively estimate the probability of the following events h 3. *Somebody you know will get infected by the virus.* 4. *Somebody you know will die because of the virus.* 5. *You will directly suffer due to the economic impact (for example run out of business, lose a job or investment).* -6. *The economy of your country will suffer badly and it will take years to come back to the pre-2020 level.* +6. *The economy of your country will suffer badly and it will take years to come back to pre-2020 level.* 7. *A single average person will get infected.* -B. *Please indicate when you think the following will happen (or have happened):* +B. *Please indicate when do you think the following will happen (or have happened):* *The end of the pandemic.* [month, year] *Everyday life comes back to normal.* [month, year] *Do you think the pandemic will comeback in a second wave? If so, when?* [yes/no] [month, year] -- GitLab From f37ef7941fa708519a1380aec7dbf5052e011312 Mon Sep 17 00:00:00 2001 From: Ondrej Zika Date: Fri, 27 Mar 2020 12:36:31 +0100 Subject: [PATCH 2/2] Update proposal.md --- docs/proposal.md | 12 +----------- 1 file changed, 1 insertion(+), 11 deletions(-) diff --git a/docs/proposal.md b/docs/proposal.md index c34fdf9..2ef9f51 100755 --- a/docs/proposal.md +++ b/docs/proposal.md @@ -4,16 +4,9 @@ In a series of laboratory studies investigating the role of trait anxiety in ave *Fig. 1: Data show trial-by-trial adjustment of shock expectancy after a switch from high to low (blue) and low to high (red) probability of shock. Lighter colors show high trait anxious group.* ![behavioural_data](behavioural_data.png) -(Nir: i would take the legend and put it under the two subplots, otherwise mainly because there is no differnece in the reds on the left, its a bit confusing) - The above findings may potentially explain why we see higher rates of fear relapse in highly anxious individuals. Our group is currently conducting a study investigating whether state switching and high trait anxiety also lead to higher rates of relapse on a session one week later. The ongoing global crisis related to the COVID-19 pandemic resembles our experimental design in a number of ways: There is an initial fear period which will be followed by a phase of fear extinction and most likely (see below) by a period of fear reinstatement. We therefore see the situation as a unique opportunity to investigate fear learning in a natural setting. -(Nir: i would seperate here between what is clear and what is prediction to future re the corona virus) - -Nir: im missing here your hypothesis in clear words, something like: -based on the predicted number of corona cases (X et al) we expect a general corresponding increase in anxiety charecterized by a temporal delay (Nir: bad phrasing but you get the point) -Based on our previous finidngs, we thus expect that high anxoues would show even more..... compared to low anxouse.... ### Hypotheses @@ -83,10 +76,7 @@ As alternative, to motivate people a bit more to complete, we can raise interim #### Questions -This section contains all questionnaires newly designed for this study. - -(Nir: make clear what will the participants will see and what is explanations for the researcher/ethics etc..) - +This section contains all questionnaires newly designed for this study. Information about purpose and responses won't be seen to participants. --- -- GitLab