complete the assignment Write a 2 page paper (2 pages of TEXT, 1 inch margins, Times New Roman 12 point font, double spaced. You do not need an abstract or a title page) summarizing what the researchers did, what they found to be significant and interesting, any critical evaluations of their methods or findings, and your personal response to the article. The paper should be in your own words. You should not cut and paste any content from the article. Send your 2-page paper as a .docx or .doc file

I’m working on a psychology writing question and need support to help me study.

 

Get Your Custom Essay Written From Scratch
We have worked on a similar problem. If you need help click order now button and submit your assignment instructions.
Just from $9/Page
Order Now

complete the assignment Write a 2 page paper (2 pages of TEXT, 1 inch margins, Times New Roman 12 point font, double spaced. You do not need an abstract or a title page) summarizing what the researchers did, what they found to be significant and interesting, any critical evaluations of their methods or findings, and your personal response to the article. The paper should be in your own words. You should not cut and paste any content from the article. Send your 2-page paper as a .docx or .doc file

 

UNFORMATTED ATTACHMENT PREVIEW

835462 research-article2019 PSSXXX10.1177/0956797619835462Ballard et al.Information Processing Under Reward Versus Under Punishment ASSOCIATION FOR Research Article PSYCHOLOGICAL SCIENCE Information Processing Under Reward Versus Under Punishment Psychological Science 2019, Vol. 30(5) 757­–764 © The Author(s) 2019 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/0956797619835462 https://doi.org/10.1177/0956797619835462 www.psychologicalscience.org/PS Timothy Ballard, David K. Sewell, Daniel Cosgrove, and Andrew Neal School of Psychology, The University of Queensland Abstract Much is known about the effects of reward and punishment on behavior, yet little research has considered how these incentives influence the information-processing dynamics that underlie decision making. We fitted the linear ballistic accumulator to data from a perceptual-judgment task to examine the impacts of reward- and punishmentbased incentives on three distinct components of information processing: the quality of the information processed, the quantity of that information, and the decision threshold. The threat of punishment lowered the average quality and quantity of information processed, compared with the prospect of reward or no performance incentive at all. The threat of punishment also induced less cautious decision making by lowering people’s decision thresholds relative to the prospect of reward. These findings suggest that information-processing dynamics are determined not only by objective properties of the decision environment but also by the higher order goals of the system. Keywords information processing, reward, punishment, decision making, computational modeling, open data Received 4/8/18; Revision accepted 12/27/18 One of the most fundamental drivers of human behavior is our motivation to obtain rewards and avoid punishment. Sensitivity to pleasure and pain has been critical for our survival (Leknes & Tracey, 2008). Animals that could successfully respond to rewarding stimuli (e.g., mates) and evade threatening stimuli (e.g., predators) stood a greater chance of passing on their genes. Motivation for seeking reward and avoiding punishment may also be critical for shaping decision making on a more moment-to-moment basis, as people respond differently to situations in which they are rewarded for success versus punished for failure (Frank, Seeberger, & O’Reilly, 2004). Although much is known about how reward and punishment influence motivation and behavior, less is known about their effects on decision making. Contemporary models of decision making assume that people process information by integrating information sampled from the environment over time until it provides sufficient evidence for an action (Brown & Heathcote, 2008; Ratcliff & Smith, 2004; Sewell & Smith, 2016). In this research, we considered the effects of reward and punishment on three components of information processing: (a) the average quantity of task-relevant information that is sampled in a given unit of time, (b) the quality or diagnostic value of task-relevant information that is sampled, and (c) the decision threshold that determines when to act on that information. We used a well-validated computational model—the linear ballistic accumulator (LBA)—to quantify these components of the information-processing system. The LBA (Brown & Heathcote, 2008) assumes that evidence is accumulated for each response alternative independently until the evidence for one alternative breaches a threshold, at which point that alternative is selected. The LBA estimates a mean drift-rate parameter for each response alternative, which summarizes the amount of evidence accumulated for that alternative per unit of time. The difference in the mean drift rates between Corresponding Author: Timothy Ballard, The University of Queensland, School of Psychology, McElwain Building, St. Lucia, QLD 4072, Australia E-mail: t.ballard@uq.edu.au Ballard et al. 758 the correct and incorrect alternatives summarizes the discriminative evidence driving the decision process, capturing the quality of the information being processed. A novel aspect of this research is our use of the LBA to measure the quantity of information processed, which can be indexed by taking the sum of the drift rates for the correct and incorrect responses. The LBA also estimates a threshold parameter for each response alternative, which quantifies the amount of evidence required to select that alternative, and reflects the cautiousness of the decision maker. By examining how these parameters are affected by reward and punishment, we can make inferences about how these factors might influence the information-processing dynamics that underlie decision making. There are a number of ways that reward and punishment might influence information processing. One possibility is that people may invest more effort when seeking reward or avoiding punishment (Brehm & Self, 1989; Kruglanski et al., 2012). Many researchers have argued that the application of mental effort is costly and that people will exert effort only if the benefits outweigh the costs (e.g., Kahneman, 1973; Kool & Botvinick, 2014). The effect of increasing the amount of effort that is allocated to a task is to enhance the informationprocessing capacity for that task. Additional processing capacity could influence the decision process in multiple ways. First, it might improve the ability of the decision maker to discriminate between competing response alternatives, which would correspond to an increase in the quality of the information extracted from the environment. Second, it might enable the decision maker to process information more rapidly without necessarily affecting the ability to discriminate between alternatives. This would correspond to an increase in the quantity of the information that is processed per unit of time. The quality and quantity of information processing are theoretically distinct. Yet researchers who have used sequential-sampling models such as the LBA to examine the mechanisms underlying decision making have focused almost exclusively on the quality component, with no research (to our knowledge) explicitly disentangling the two constructs. Thus, it is unclear whether reward and punishment should affect quality, quantity, both, or neither. A competing perspective is that the presence of reward or punishment may induce performance monitoring, which diverts resources away from the focal task (e.g., Hockey, 1997; Kanfer & Ackerman, 1989). For example, Kanfer and Ackerman (1989) argued that when people are given a goal to achieve, they actively monitor their performance, and this can impair performance because it consumes resources. Under this view, reward and punishment may hinder the ability to process task-related information if the performance-monitoring costs cannot be offset by increases in processing capacity. This account makes competing predictions to the capacity account described above. These arguments suggest that reward and punishment may decrease the amount of resources that are available for the task itself, which could reduce the quality or quantity of the information that is extracted from the environment. There is also reason to believe that the effects of punishment may be different from the effects of reward. Research on loss aversion suggests that people are more motivated to avoid losses than to obtain gains of equivalent magnitude (Kahneman & Tversky, 1979). This may lead people to apply more effort when faced with punishment compared with rewards, leading to enhanced capacity under punishment. In this case, the quality or quantity of information processing should be greater under punishment than under reward. However, loss aversion might also increase the tendency to monitor one’s performance when under threat of punishment, which would detract from the resources available for the task itself. In this case, the quality or quantity of information processing should be greater under reward than under punishment. It is also possible that both of these effects emerge, in which case the difference in quality and quantity between reward and punishment would depend on the extent to which increases in capacity due to effort are offset by performance-monitoring costs. Finally, it is possible that reward and punishment influence not only the quality or quantity of information processing but also the threshold that determines the amount of information required before an action is enacted. Regulatory-focus theory (Higgins, 1997) proposes that people are more cautious when focused on preventing negative outcomes compared with when pursuing positive outcomes. This suggests an asymmetry in the amount of information that people process before being willing to commit to a course of action. Specifically, regulatory-focus theory suggests that people may require less information and, thus, set a lower response threshold when seeking reward than when avoiding punishment. In the following sections, we report an experiment in which we examined these processes using a wellestablished perceptual-judgment paradigm. We manipulated reward and punishment within participants and used the LBA to measure quality and quantity of information processed and the decision threshold in each condition. Method Thirty-five undergraduate students from The University of Queensland (23 female, 12 male; mean age = 19.89 years) performed a random-dot-motion discrimination task. In each trial, participants were presented with a Information Processing Under Reward Versus Under Punishment 759 Blank Screen (500 ms) Fixation Cross (500 ms) Stimulus Presentation (up to 2,000 ms) Blank Screen Before Next Trial (500 ms) Fig. 1. Sequence of an experimental trial. During each stimulus presentation, some proportion of the 40 dots on screen moved coherently toward either the top left or the top right of the cloud, while the other dots moved randomly. Participants had to identify whether the dots were moving mostly to the left or mostly to the right. (The stimulus image shown here is not drawn to scale.) cloud of 40 moving dots (see Fig. 1). Across successive frames during stimulus presentation, some proportion of the dots moved coherently toward either the top left or the top right of the cloud, while the other dots moved randomly. The participant’s task was to identify whether the dots were moving mostly to the left or mostly to the right. Difficulty was manipulated by varying the proportion of the dots moving coherently (.08 = easy, .06 = medium, .05 = hard). The stimulus was generated using a version of the white-noise algorithm (Pilly & Seitz, 2009). The diameter of the cloud stimulus was approximately 33% of the screen height. Each dot had a diameter of approximately 0.66% of the screen height. The dot positions were updated approximately every 66.6 ms. On each update, dots were independently selected according to the trial coherence level to move coherently or randomly on the next update. Coherent dots moved at a rate such that it would take a single dot 3,000 ms to traverse the entire cloud. If the new location would be outside the cloud, the dot was repositioned randomly. Noncoherent dots were repositioned randomly. At the start of the experiment, participants were instructed to press the “z” key if the dots were streaming up to the left and the “/” key if they were streaming up to the right. They then completed 10 practice trials. The experiment was broken down into nine blocks of 90 trials each. For each block, participants were given the goal of achieving 77% accuracy and a mean response time of less than 1.071 s (these values were based on the accuracy and mean response time observed in a pilot study). Participants started the experiment with an endowment of $7.50 (AUD). Reward and punishment were manipulated within participants by varying the monetary incentive associated with the block (for similar manipulations, see Galea, Mallia, Rothwell, & Diedrichsen, 2015; Guitart-Masip et al., 2012; O’Doherty, Kringlebach, Rolls, Hornak, & Andrews, 2001; Palminteri, Khamassi, Joffily, & Coricelli, 2015). In one third of blocks, participants gained $2.50 when they achieved the goal (the reward condition). In another third, participants lost $2.50 when they did not achieve the goal (and did not gain any money when they achieved it; the punishment condition). In the other third, they could not gain or lose any money regardless of whether they achieved the goal (the neutral condition). The block incentives occurred in random order but were stratified such that each condition occurred once in Blocks 1 to 3, once in Blocks 4 to 6, and once in Blocks 7 to 9. At the start of each block, participants were reminded of the goal and were told the consequences of achieving or not achieving the goal in that block. Within each block, half of the stimuli displayed dots moving up and to the left and the other half displayed dots moving Ballard et al. 760 950 Mean Response Time (ms) Percentage Correct 75 70 65 60 55 900 850 800 Neutral Reward Punishment Easy Neutral Medium Reward Punishment Hard Fig. 2. Percentage correct and mean response time as a function of difficulty in the neutral, reward, and punishment conditions. Error bars indicate ±1 SE. up and to the right. Difficulty was also manipulated within blocks by varying the level of coherent motion across trials. Stimuli were presented in random order. Each trial terminated when the participant made a response or after 2 s, whichever came first. If the participant did not make a response within 2 s, the words “TOO SLOW” appeared, and the trial was recorded as a nonresponse. At the end of the block, the participant received feedback regarding whether the block goal had been achieved. However, participants did not receive trial-by-trial feedback regarding the correctness of each decision. The resulting sample size was 28,170 total experimental trials.1 This is consistent with the recommended sample size for accurate parameter estimation using the LBA (e.g., Donkin, Averell, Brown, & Heathcote, 2009). We excluded from the analysis all practice trials, nonresponses, and trials in which the observed response time was 250 ms or less. These response times are so short that the participant could not have plausibly made a decision during this time. The excluded cases made up slightly more than 3% of all trials. The data and code necessary to conduct the analyses as well as the source code used to generate the experimental stimuli are publicly available and can be found on the Open Science Framework at osf.io/kwvu6. Results We first considered the effects of the experimental manipulations on accuracy and response time (see Fig. 2). We analyzed these effects using Bayesian methods (full details regarding the analyses are included in the Supplemental Material available online). We report the Bayes factor (BF) as a measure of the evidence for each effect, using the heuristic classification scheme reported by Lee and Wagenmakers (2013) to describe the strength of evidence. We also report the 95% credible interval (CI) on the estimate to evaluate the magnitude of each effect. When people were rewarded for goal achievement, we found moderate evidence for no change in accuracy when compared with people who received no incentive (i.e., the neutral condition; BF = 0.01, 95% CI = [−0.03, 0.10]). We also found anecdotal evidence for no difference in response time between the reward and neutral conditions (BF = 0.39, 95% CI = [−7.30, 9.09]). When participants were punished for goal failure, we found extreme evidence for a difference in accuracy relative to the neutral condition, with accuracy being lower under punishment (BF = 184.06, 95% CI = [−0.24, −0.11]). We also found moderate evidence for a difference in response time between the punishment and neutral conditions, with participants taking longer to respond under punishment (BF = 4.09, 95% CI = [1.13, 18.58]). We found very strong evidence for an effect of difficulty on accuracy, with participants responding less accurately as difficulty increased (BF = 92.77, 95% CI = [−0.20, −0.14]). We also found extreme evidence for an effect of difficulty on response time, with participants taking longer to respond as difficulty increased (BF = 9,076.23, 95% CI = [15.49, 25.25]). As can be seen in Figure 2, when participants were threatened with punishment, they responded about as accurately (and no faster) to easy stimuli as they did to medium-difficulty stimuli when either rewarded or not incentivized. Likewise, they responded about as Information Processing Under Reward Versus Under Punishment accurately (and no faster) to medium-difficulty stimuli under punishment as they did to hard stimuli under reward. These results point to a considerable performance cost associated with the threat of punishment. Although the behavioral results are suggestive, we cannot make inferences about the underlying decision process on the basis of these results alone. Specifically, we cannot determine the extent to which the effects on accuracy and response times are due to changes in the quality of information that people process under different incentives, the quantity of information being processed, or changes in decision thresholds. To examine the underlying information-processing dynamics, we Quality 761 fitted the choice and response time data using the LBA model within a hierarchical Bayesian framework. We report the results of this analysis in the next section. Quality and quantity of information processed Using the measures described above, we first examined the quality and quantity of the information being processed in the reward, punishment, and neutral conditions (see Fig. 3). We found extreme evidence for a difference in the quality of information processed, which reflects the ability to discriminate between rightward and leftward Quantity Threshold 5.4 0.9 Mean Value 1.2 5.2 0.8 0.7 5.0 0.6 1.1 4.8 1.0 0.5 Neutral Reward Punishment Neutral Participant Posterior Mean Quality Reward Punishment Neutral Quantity Reward Punishment Threshold 1 0 −1 −2 Neutral Reward Punishment Neutral Reward Punishment Neutral Reward Punishment Fig. 3. Quality and quantity of information processing and the decision threshold in the neutral, reward, and punishment conditions. Quality was determined by the difference in mean drift rates. Quantity was determined by the sum of the mean drift rates. Quality and quantity were collapsed across difficulty. The decision threshold was collapsed across response. The violin plots in the top row represent the posterior density of the condition mean. The bottom row displays posterior means for each participant, which were normalized by subtracting the value in each condition from the corresponding value in the neutral condition. The blue lines in the bottom row indicate participants for whom the relevant component had a higher value in the reward condition than in the punishment condition. The red lines indicate participants for whom the component had a higher value in the punishment condition than in the reward condition. Ballard et al. 762 motion, between the reward and punishment conditions. Quality was higher in the reward condition than in the punishment condition (BF = 15,423, 95% CI = [0.18, 0.32]). We also found extreme evidence for a difference in the quality of information processed between the punishment and neutral conditions, with the quality being higher in the neutral condition than in the punishment condition (BF = 15,466, 95% CI = [0.13, 0.27]). Finally, we found moderate evidence for no difference in the quality of information processed between the reward and neutral conditions (BF = 0.19, 95% CI = [−0.03, 0.11]). With regard …
Purchase answer to see full attachment
Student has agreed that all tutoring, explanations, and answers provided by the tutor will be used to help in the learning process and in accordance with Studypool’s honor code & terms of service.
Do you need a similar assignment done for you from scratch? We have qualified writers to help you
Use our paper writing service to score better and meet your deadlines.
Order Now