Limits...
Decision making from economic and signal detection perspectives: development of an integrated framework.

Lynn SK, Wormwood JB, Barrett LF, Quigley KS - Front Psychol (2015)

Bottom Line: Conversely, traditional signal detection approaches to decision making focus on the uncertainty that arises from variability in perceptual signals and typically ignore the influence of outcome value uncertainty.Here, we compare and contrast the economic and signals frameworks that guide research in decision making, with the aim of promoting their integration.We show that an integrated framework can expand our ability to understand a wider variety of decision-making behaviors, in particular the complexly determined real-world decisions we all make every day.

View Article: PubMed Central - PubMed

Affiliation: Interdisciplinary Affective Science Laboratory, Department of Psychology, Northeastern University Boston, MA, USA.

ABSTRACT
Behavior is comprised of decisions made from moment to moment (i.e., to respond one way or another). Often, the decision maker cannot be certain of the value to be accrued from the decision (i.e., the outcome value). Decisions made under outcome value uncertainty form the basis of the economic framework of decision making. Behavior is also based on perception-perception of the external physical world and of the internal bodily milieu, which both provide cues that guide decision making. These perceptual signals are also often uncertain: another person's scowling facial expression may indicate threat or intense concentration, alternatives that require different responses from the perceiver. Decisions made under perceptual uncertainty form the basis of the signals framework of decision making. Traditional behavioral economic approaches to decision making focus on the uncertainty that comes from variability in possible outcome values, and typically ignore the influence of perceptual uncertainty. Conversely, traditional signal detection approaches to decision making focus on the uncertainty that arises from variability in perceptual signals and typically ignore the influence of outcome value uncertainty. Here, we compare and contrast the economic and signals frameworks that guide research in decision making, with the aim of promoting their integration. We show that an integrated framework can expand our ability to understand a wider variety of decision-making behaviors, in particular the complexly determined real-world decisions we all make every day.

No MeSH data available.


Elements of the signals framework (after Lynn and Barrett, 2014). In emotion perception, for example, facial expressions are evaluated by one person (the perceiver) to determine the emotional state of another person (the sender). Signals, depicted on the x-axis, comprise two categories: targets (defining, e.g., what the sender looks like when she is angry) and foils (defining, e.g., what the sender looks like when she is not angry). Signals from either category vary over a perceptual domain such as “scowl intensity.” Any signal (i.e., a particular scowl intensity) can arise from either category, with a likelihood given by the target and foil distributions. Perceivers therefore, experience uncertainty about the category membership of any particular signal. Here, the perceiver responds to facial expressions to the right of criterion (red arrow) as if they were angry, and to facial expressions to the left of criterion as if they were not angry. Perceivers make a decision between two options and the perceptual uncertainty yields four possible outcomes: (1) Classifying a stimulus as a target when it is a target is a correct detection. (2) Classifying a stimulus as a target when it is a foil is a false alarm. (3) Classifying a stimulus as a foil when it is a target is a missed detection. (4) Classifying a stimulus as a foil when it is a foil is a correct rejection. Measures of sensitivity (e.g., d') characterize perceptual uncertainty, depicted here as overlap of the target and foil distributions. Measures of bias (e.g., c or beta) characterize the decision criterion's location in the perceptual domain. Sensitivity and bias are derived from the numbers of correct detections and false alarms committed over a series of decisions (Macmillan and Creelman, 2005). Perceptual uncertainty causes the perceiver to make mistakes regardless of his or her degree of bias; missed detections cannot be reduced without increasing false alarms.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4495727&req=5

Figure 4: Elements of the signals framework (after Lynn and Barrett, 2014). In emotion perception, for example, facial expressions are evaluated by one person (the perceiver) to determine the emotional state of another person (the sender). Signals, depicted on the x-axis, comprise two categories: targets (defining, e.g., what the sender looks like when she is angry) and foils (defining, e.g., what the sender looks like when she is not angry). Signals from either category vary over a perceptual domain such as “scowl intensity.” Any signal (i.e., a particular scowl intensity) can arise from either category, with a likelihood given by the target and foil distributions. Perceivers therefore, experience uncertainty about the category membership of any particular signal. Here, the perceiver responds to facial expressions to the right of criterion (red arrow) as if they were angry, and to facial expressions to the left of criterion as if they were not angry. Perceivers make a decision between two options and the perceptual uncertainty yields four possible outcomes: (1) Classifying a stimulus as a target when it is a target is a correct detection. (2) Classifying a stimulus as a target when it is a foil is a false alarm. (3) Classifying a stimulus as a foil when it is a target is a missed detection. (4) Classifying a stimulus as a foil when it is a foil is a correct rejection. Measures of sensitivity (e.g., d') characterize perceptual uncertainty, depicted here as overlap of the target and foil distributions. Measures of bias (e.g., c or beta) characterize the decision criterion's location in the perceptual domain. Sensitivity and bias are derived from the numbers of correct detections and false alarms committed over a series of decisions (Macmillan and Creelman, 2005). Perceptual uncertainty causes the perceiver to make mistakes regardless of his or her degree of bias; missed detections cannot be reduced without increasing false alarms.

Mentions: In the signals framework (Green and Swets, 1966; Lynn and Barrett, 2014), a perceiver classifies a percept as one of two possible options. In a detection decision, the perceiver decides if the percept was present (i.e., the “signal” option) or absent (i.e., the “noise” option). For example, a perceiver might be asked to determine whether a tone is present or not on each trial of an auditory task in which some trials contain non-tone white noise and other trials contain a tone of variable acoustic intensity embedded in white noise. In an identification decision, the perceiver decides if the percept (a “signal”) is an exemplar of one category (Option A) or another (Option B). For example, on each trial of an emotion perception task, a perceiver might be asked to determine whether an ambiguous facial expression communicates a state of happiness or anger. In both detection and identification paradigms, the two options span different but overlapping ranges of a single, continuous, perceptual dimension, such as decibels of acoustic energy or expressive facial features1 (Figure 4). Thus, the signals framework focuses on variability in the perceptual features of the options (e.g., how similar in appearance exemplars from Option A are to those of Option B). We will frame our discussion of the signals framework in terms of identification paradigms, and not further distinguish between detection and identification here [but see Green and Swets (1966) and Macmillan and Creelman (2005), for more details on this distinction].


Decision making from economic and signal detection perspectives: development of an integrated framework.

Lynn SK, Wormwood JB, Barrett LF, Quigley KS - Front Psychol (2015)

Elements of the signals framework (after Lynn and Barrett, 2014). In emotion perception, for example, facial expressions are evaluated by one person (the perceiver) to determine the emotional state of another person (the sender). Signals, depicted on the x-axis, comprise two categories: targets (defining, e.g., what the sender looks like when she is angry) and foils (defining, e.g., what the sender looks like when she is not angry). Signals from either category vary over a perceptual domain such as “scowl intensity.” Any signal (i.e., a particular scowl intensity) can arise from either category, with a likelihood given by the target and foil distributions. Perceivers therefore, experience uncertainty about the category membership of any particular signal. Here, the perceiver responds to facial expressions to the right of criterion (red arrow) as if they were angry, and to facial expressions to the left of criterion as if they were not angry. Perceivers make a decision between two options and the perceptual uncertainty yields four possible outcomes: (1) Classifying a stimulus as a target when it is a target is a correct detection. (2) Classifying a stimulus as a target when it is a foil is a false alarm. (3) Classifying a stimulus as a foil when it is a target is a missed detection. (4) Classifying a stimulus as a foil when it is a foil is a correct rejection. Measures of sensitivity (e.g., d') characterize perceptual uncertainty, depicted here as overlap of the target and foil distributions. Measures of bias (e.g., c or beta) characterize the decision criterion's location in the perceptual domain. Sensitivity and bias are derived from the numbers of correct detections and false alarms committed over a series of decisions (Macmillan and Creelman, 2005). Perceptual uncertainty causes the perceiver to make mistakes regardless of his or her degree of bias; missed detections cannot be reduced without increasing false alarms.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4495727&req=5

Figure 4: Elements of the signals framework (after Lynn and Barrett, 2014). In emotion perception, for example, facial expressions are evaluated by one person (the perceiver) to determine the emotional state of another person (the sender). Signals, depicted on the x-axis, comprise two categories: targets (defining, e.g., what the sender looks like when she is angry) and foils (defining, e.g., what the sender looks like when she is not angry). Signals from either category vary over a perceptual domain such as “scowl intensity.” Any signal (i.e., a particular scowl intensity) can arise from either category, with a likelihood given by the target and foil distributions. Perceivers therefore, experience uncertainty about the category membership of any particular signal. Here, the perceiver responds to facial expressions to the right of criterion (red arrow) as if they were angry, and to facial expressions to the left of criterion as if they were not angry. Perceivers make a decision between two options and the perceptual uncertainty yields four possible outcomes: (1) Classifying a stimulus as a target when it is a target is a correct detection. (2) Classifying a stimulus as a target when it is a foil is a false alarm. (3) Classifying a stimulus as a foil when it is a target is a missed detection. (4) Classifying a stimulus as a foil when it is a foil is a correct rejection. Measures of sensitivity (e.g., d') characterize perceptual uncertainty, depicted here as overlap of the target and foil distributions. Measures of bias (e.g., c or beta) characterize the decision criterion's location in the perceptual domain. Sensitivity and bias are derived from the numbers of correct detections and false alarms committed over a series of decisions (Macmillan and Creelman, 2005). Perceptual uncertainty causes the perceiver to make mistakes regardless of his or her degree of bias; missed detections cannot be reduced without increasing false alarms.
Mentions: In the signals framework (Green and Swets, 1966; Lynn and Barrett, 2014), a perceiver classifies a percept as one of two possible options. In a detection decision, the perceiver decides if the percept was present (i.e., the “signal” option) or absent (i.e., the “noise” option). For example, a perceiver might be asked to determine whether a tone is present or not on each trial of an auditory task in which some trials contain non-tone white noise and other trials contain a tone of variable acoustic intensity embedded in white noise. In an identification decision, the perceiver decides if the percept (a “signal”) is an exemplar of one category (Option A) or another (Option B). For example, on each trial of an emotion perception task, a perceiver might be asked to determine whether an ambiguous facial expression communicates a state of happiness or anger. In both detection and identification paradigms, the two options span different but overlapping ranges of a single, continuous, perceptual dimension, such as decibels of acoustic energy or expressive facial features1 (Figure 4). Thus, the signals framework focuses on variability in the perceptual features of the options (e.g., how similar in appearance exemplars from Option A are to those of Option B). We will frame our discussion of the signals framework in terms of identification paradigms, and not further distinguish between detection and identification here [but see Green and Swets (1966) and Macmillan and Creelman (2005), for more details on this distinction].

Bottom Line: Conversely, traditional signal detection approaches to decision making focus on the uncertainty that arises from variability in perceptual signals and typically ignore the influence of outcome value uncertainty.Here, we compare and contrast the economic and signals frameworks that guide research in decision making, with the aim of promoting their integration.We show that an integrated framework can expand our ability to understand a wider variety of decision-making behaviors, in particular the complexly determined real-world decisions we all make every day.

View Article: PubMed Central - PubMed

Affiliation: Interdisciplinary Affective Science Laboratory, Department of Psychology, Northeastern University Boston, MA, USA.

ABSTRACT
Behavior is comprised of decisions made from moment to moment (i.e., to respond one way or another). Often, the decision maker cannot be certain of the value to be accrued from the decision (i.e., the outcome value). Decisions made under outcome value uncertainty form the basis of the economic framework of decision making. Behavior is also based on perception-perception of the external physical world and of the internal bodily milieu, which both provide cues that guide decision making. These perceptual signals are also often uncertain: another person's scowling facial expression may indicate threat or intense concentration, alternatives that require different responses from the perceiver. Decisions made under perceptual uncertainty form the basis of the signals framework of decision making. Traditional behavioral economic approaches to decision making focus on the uncertainty that comes from variability in possible outcome values, and typically ignore the influence of perceptual uncertainty. Conversely, traditional signal detection approaches to decision making focus on the uncertainty that arises from variability in perceptual signals and typically ignore the influence of outcome value uncertainty. Here, we compare and contrast the economic and signals frameworks that guide research in decision making, with the aim of promoting their integration. We show that an integrated framework can expand our ability to understand a wider variety of decision-making behaviors, in particular the complexly determined real-world decisions we all make every day.

No MeSH data available.