تصمیم‌گیری ادراکی در موقعیت‌های واقع‌گرایانه با تلفیق شواهد بصری و نشانه‌های اجتماعی: مدل رانش انتشار به ‌عنوان ابزار تحلیلی

مقدمه: در موقعیت‌های واقع‌گرایانه، تصمیم‌گیری‌ها عمدتاً بر اساس ترکیبی از اطلاعاتی که از منابع مختلف به دست می‌آیند، صورت می‌گیرد. هدف اصلی این مطالعه، بررسی تاثیرات اطلاعات اجتماعی بر تصمیم‌گیری در حضور شواهد حسی، و مطالعه نحوه وقوع این تاثیرات بود.روش کار: در پژوهش حاضر با استفاده از آزمون تصمیم‌ گیری ادراکی، آزمایشی در مورد تصمیم ‌گیری در حضور شواهد حسی و اجتماعی طراحی شد، تا با استفاده از آن، تاثیرات اجتماعی هنجاری به حداقل رسد و تنها بر تاثیرات اجتماعی اطلاعاتی تمرکز شود. معناداری تحلیل ‌های رفتاری با استفاده از آزمون ‌های آماری انجام شد و مدل رانش انتشار تعمیم ‌یافته برای بررسی چگونگی فرایند تصمیم ‌گیری استفاده شد. یافته‌ها: نتایج نشان می‌دهند که اطلاعات اجتماعی به طور قابل‌توجهی دقت تصمیم‌گیری شرکت‌کنندگان را افزایش می‌دهد، حتی زمانی که شرکت‌کنندگان اطلاعات را از شرکای با دقت پایین دریافت می‌کنند. افزایش صحت شرکت‌کننده به طرز چشمگیری با میزان درستی اطلاعات اجتماعی دریافت ‌شده (از شریک با صحت بالاتر نسبت به شریک با صحت پایین‌تر) افزایش می‌یابد، در حالی‌ که قطعیت شریکان تاثیر چندانی روی این بهبود نمی‌گذارد. در آنالیزهای مربوط به مدل رانش انتشار از مقایسه داده‌ های مربوط به آزمایش مرجع و داده ‌های همراه با اطلاعات اجتماعی، در یافتیم تمام پارامترهای تصمیم‌گیری از جمله پارامترهای آستانه تصمیم، نقطه شروع و نرخ رانش به صورت معناداری متفاوت هستند و آستانه تصمیم و نقطه شروع در آزمایشات همراه با اطلاعات اجتماعی بیشتر از حالت مرجع و نرخ رانش کمتر از حالت مرجع است، اما تفاوت‌های دقت میان مواجهه با شرکای با دقت بالا و پایین عمدتاً تحت‌تاثیر پارامتر نقطه شروع قرار دارند.نتیجه‌گیری: در آخر، این مطالعه در پژوهش‌های مرتبط با تصمیم‌گیری، رفتار انسانی و بهبود کیفیت تصمیم‌گیری‌ها مفید است و به ما در درک بهتر اثر اطلاعات اجتماعی بر فرایند تصمیم‌گیری افراد کمک می‌کند.  

perceptual decision-making in realistic situations with the integration of visual evidence and social cues: the diffusion model as an analytical tool

perceptual decision-making is a pivotal aspect of our daily lives, as various decision-making processes involve accumulating evidence from sensory perception, personal preferences, social cues, and other sources. in the realm of social psychology, the impact of social information on decision-making is a subject of interest, studied under the umbrella of social influence. social influence involves changes in cognition, attitudes, or behavior due to external factors (1), and conformity refers to the frequency with which individuals mimic social cues, regardless of their accuracy. social influence arises from two primary motivations: informational influence, where individuals seek group-derived information because they believe it to be more accurate, and normative influence, driven by the desire to align with group norms (2, 3). well-designed experiments are essential to draw meaningful conclusions and disentangle normative and informational influences. classic conformity studies have highlighted the challenges associated with both types of social influence. it is necessary to distinguish between them to ensure the interpretation of research findings, obstructing a comprehensive understanding of decision-making in social contexts.this research focuses on minimizing the factors producing normative effects and investigating the impact of informational social influence on decision-making when social information is presented after distinct sensory cues. the present study aimed  to understand how last-minute social information may impact the sensory gathering process and, subsequently, influence decision-making.methodsthe experimental task in this study comprised six stages, with five involving extensive training, followed by the main task. the main experiment was divided into four sessions, each consisting of one block.in this psychophysical task, participants were exposed to two 120-millisecond sensory stimuli originating from random dot motion (rdm), featuring different degrees of evidence strength and motion direction. afterward, a graphical cue was displayed, using shapes and colors familiar to participants from their training. notably, this graphical cue was solely provided to indicate the other participants’ responses, intentionally devoid of any information regarding the identity of the respondents, to minimize normative influence. they were informed that these cues represented responses from other participants who had encountered the same rdms. however, the responses were not from actual participants; instead, the researchers generated them using a signal detection model. to create the social data, the researchers relied on participant behavior in sensory trials as a foundation. they employed the model to generate responses from four participants, each with distinct performance and confidence levels. this study employed the generalized drift diffusion model (gddm) to provide a flexible framework for decision-making research. gddm allows for a broad range of experimental designs and empirical hypotheses, accommodating customized distributions for parameters such as starting position, decision threshold, and drift rate (4). this flexibility enhances our ability to investigate the influence of social information on decision processes. the present analysis required the development of customized parameters due to the unique experimental conditions, including the presentation of discrete evidence and the incorporation of social information. the model used in this study propose that the drift rate, reflecting the speed of information uptake, remains consistent throughout each trial, each consisting of two pulses. however, its effect varies between pulses depending on their coherence levels. specifically, this research derive this measure of drift rate from the initial pulse of information and compute it separately for each pulse based on its coherence level. in this manner, the researchers manage the transition between discrete pulses of information. by adopting this definition of the starting point as the ratio of the preference at time 0 to the size of the decision boundary (5) and considering that the starting point shifts to favor a more frequent or highly rewarded response when trials are manipulated (6, 7), it becomes possible to apply it to social cues. this accounts for the occurrence of a bias or preference in evidence accumulation toward social cues. however,  extending its timing to presentinga social cue is essential. by integrating this concept into the model, this study can effectively explain how social cues influence decision bias.the task is run using the psychtoolbox in matlab, and behavioral analysis takes place within matlab. statistical tests are carried out using jasp, and the gddm model is executed through the pyddm, a drift-diffusion modeling framework for python.resultsnine participants successfully completed all tasks. during the training phase, an average of 2100 data points were collected from each participant, and in the main test phase, 7200 trials were recorded. additionally, 5400 trials were recorded in the isolated task. participants notably increased their accuracy in group tasks compared to solo tasks, mainly when their partners were more precise.  however, the overall mean accuracy of participants remained significantly lower than that of their partners.the extent of adapting to social cues is another crucial factor in participant behavior. as anticipated, participants exhibited greater conformity with more accurate partners. statistical tests revealed a significant positive correlation between the influence of social partners. it was found to be significant only when comparing high-accuracy and low-accuracy partners to validate the significance of conformity behavior and identify differences between conditions. no significant difference in conformity was observed between partners with high-and low-confidence social information.the obtained findings, as modeled through ddm, reveal significant differences in crucial decision-making parameters between the isolated phase and the social task. specifically, when comparing parameters such as the drift rate (the pace of evidence accumulation), the decision threshold (the determiner of decision endpoints), and the starting point (illustrating how social information complements sensory evidence), the researchers observed an increase in the decision threshold and starting point, and a decrease in the drift rate in the social task compared to the isolated task. further investigation into variations among partners reveals that the disparity in accuracy improvement between high and low-accuracy partners when social information is introduced at the final stage leads to an increase in the starting point.conclusionthis study delves into the intricate dynamics of decision-making in social contexts, specifically focusing  on the impact of social information presented after distinct sensory cues. the pursuit of understanding how last-minute social information affects the sensory gathering process and, consequently, influences decision-making has unveiled valuable insights into the multifaceted world of social influence on human behavior.the obtained results shed light on the role of social cues in enhancing decision accuracy. participants exhibited a significant improvement in accuracy when operating in social blocks as opposed to isolated ones, primarily when their partners exhibited higher accuracy. nevertheless, the mean accuracy of participants remained persistently lower than that of their high-