Table 1.
Characteristics of the interpretation bias paradigms included in the review.
| Paradigm | Brief description [response format] | Strengths | Limitations |
|---|---|---|---|
| SINGLE AMBIGUOUS WORDS | |||
| Homophone task | Spoken homophones with threatening and neutral associations (e.g., pain/pane) are presented, which participants write down [written response] | Simple to administer without use of a computer | Threatening and neutral associations often have different written and verbal frequencies of use. Often a small number of appropriate stimuli |
| Homophone task with pictorial responses | Spoken homophones with threatening and neutral associations are presented, followed by pictures corresponding to each interpretation. Participants select the picture best matching the word [verbal response or pointing toward appropriate picture] | Individual differences in reading and writing ability do not affect participant responses. Appropriate for children | Threatening and neutral associations often have different written and verbal frequencies of use. Often a small number of appropriate stimuli |
| Lexical decision task | Homographs are presented individually as primes, followed by a target word (homophones may also be presented aurally). Participants indicate whether the target is a real-word or not; faster response times to negative compared to neutral target words indicate a priming effect for that meaning [keyboard or response box] | Measure of response times unlikely to be influenced by demand characteristics | Often a small number of appropriate stimuli. Low frequency target words may be subject to guessed responses |
| Homographic response task - Single word response |
Homographs are presented individually, and participants are required to write the first related word they think of [written response] | Simple to administer without use of a computer | Responses may be subject to demand characteristics. Homographic associations vary in their dominance, potentially influencing responses |
| Homographic response task - Multiple word response |
Homographs are presented individually, and participants are required to write multiple associated words [written response] | Simple to administer without use of a computer | Responses may be subject to demand characteristics. Homographic associations vary in their dominance, potentially influencing responses |
| Sentence generation task | Homographs are presented individually, and participants are required to form a sentence featuring each homograph [written response] | Simple to administer without use of a computer | Responses may be subject to demand characteristics |
| Word stem completion task | Participants are presented with word stems, which they complete with the first word coming to mind [written response] | Finite number of valid responses makes it simpler to classify answers as threatening or neutral based on pre-determined lists | Variations in frequency and word length influence stem-completion responses |
| Acoustical blend of word pairs | Acoustically blended emotional and neutral words are presented that differ by one phoneme. Participants select which word they heard from two displayed choices [keyboard or response box] | A greater number of novel stimuli previously unheard by participants | The selected response may not reflect the participant's initial resolution of ambiguity |
| AMBIGUOUS IMAGES | |||
| Emotion recognition task with unaltered facial expressions | Participants classify emotional expressions presented [keyboard or mouse response] | Simple to design and administer even without the use of a computer | Prototypical expressions are easy to decode and may produce ceiling effects |
| Similarity rating task | Pairs of emotional stimuli of different intensities (e.g., moderately angry and very angry expressions) are presented, and participants rate the similarity of each pair on a numerical scale [keyboard or mouse response] | Measures relatively implicit processes and can be used with a variety of stimuli | It is difficult to ascertain precisely which features participants use to make their similarity comparisons |
| Emotion recognition task with morphed facial expressions | Participants classify emotional expressions presented. Response times may also be recorded [keyboard or mouse response] | A range of novel stimuli can be developed with different emotional intensities | At certain proportions and in certain combinations of emotions, morphed faces may appear unnatural and unlike those commonly viewed in everyday life |
| Incidental learning task | The learning phase presents facial expressions (e.g., positive, negative) in the center of the computer screen, which are predictive of the location of a subsequent target cue (i.e., negative faces predict upper targets, positive faces predict lower targets). The test phase presents neutral facial expressions followed by targets appearing with equal frequency at upper and lower locations. An interpretation bias is evident when, following neutral expressions, participants respond faster to target cues in the location predicted by specific faces. This task may also be used with morphed facial expressions during the test phase [keyboard or response box] | Measures relatively implicit processes possibly reducing demand characteristics and response biases. May also be used with other forms of ambiguous stimuli. | At certain proportions and in certain combinations of emotions, morphed faces may appear unnatural and unlike those commonly viewed in everyday life |
| Priming paradigm with morphed faces | Participants indicate the predominate emotion in morphed faces, which are presented following emotional or neutral primes (either words or images) [keyboard or response box] | Flexibility in exploring the impact of different types (e.g., linguistic or pictorial) and categories (e.g., pain-related, negative, positive) of primes and targets | At certain proportions and in certain combinations of emotions, morphed faces may appear unnatural and unlike those commonly viewed in everyday life. Prime words have different written and verbal frequencies, which should be considered |
| Priming paradigm with un-morphed faces | Cue-target pairs are presented, and participants indicate once they have determined which facial expression is presented. Speeded responses are expected when cue and target images are interpreted in the same manner [keyboard or response box] | Unlikely to be influenced by demand characteristics as participants do not explicitly state their interpretation of facial expressions. | It remains unknown how participants interpret both cue and target faces unless specifically asked |
| Rating tasks with blended faces | Different elements of the face are blended together to form ambiguous expressions (e.g., neutral eyes blended with a smiling mouth), which may be dynamically presented (i.e., an initial expression dynamically unfolds into a final expression). Participants rate each face on a specific dimension (e.g., trustworthiness) [keyboard or mouse response] | A range of static and dynamic novel stimuli can be developed from different prototypical emotional expressions | Depending on the combination of elements, blended faces may appear unnatural and unlike those commonly viewed in everyday life |
| Ambiguous visual scenes - Forced-choice paradigm |
Forced paradigm task presents an ambiguous scene, and asks participants specific questions (e.g., for social anxiety research, “Does the scene show a popular or unpopular child?”) [keyboard or response box] | Complex scenes and situations can be depicted, including interactions between individuals. Possibly increased ecological validity compared to single faces and words | Static visual scenes are not wholly representative of dynamic interactions witnessed in everyday life. Responses may be subject to demand characteristics |
| Ambiguous visual scenes - Rating task |
Following the presentation of an ambiguous visual scene, participants are asked to rate the image on a variety of dimensions, such as valence, arousal, judgment of danger, difficulty to tolerate, and how anxiety provoking the image was [keyboard or mouse response] | Complex scenes and situations can be depicted, including interactions between individuals. Possibly increased ecological validity compared to single faces and words | Static visual scenes are not wholly representative of dynamic interactions witnessed in everyday life. Responses may be subject to demand characteristics |
| Ambiguous visual scenes - Physiological arousal |
During the presentation of an ambiguous image, data on physiological arousal (e.g., skin conductance response) is collected [recording of physiological arousal] | Complex scenes and situations can be depicted, including interactions between individuals. Possibly increased ecological validity compared to single faces and words. More objective response compared to subjective ratings | Static visual scenes are not wholly representative of dynamic interactions witnessed in everyday life. Verification needed as to whether physiological arousal truly does provide a valid and replicable measure of ambiguity resolution |
| AMBIGUOUS SCENARIOS | |||
| Ambiguous scenarios test - Provide first explanation thought of |
Following the presentation of an ambiguous scenario: - Participants write down the first explanation they think of [written response] |
Open-ended responses do not constrain participants to a set of pre-determined interpretations | Responses may be subject to demand characteristics |
| - Imagine the outcome/state what is happening | - Participants write down the first explanation they think of [written response] | Open-ended responses do not constrain participants to a set of pre-determined interpretations | Responses may be subject to demand characteristics |
| - Rank order likelihood of possible interpretations | - Participants rank order a number of presented interpretations by their likelihood of coming to mind [written response] | Simple to administer without use of a computer | Closed response format limits participants to ranking pre-determined interpretations, none of which may reflect their own interpretation |
| - Rate level of agreement with possible interpretations | - Participants rate their level of agreement of number of presented interpretations [written response] | Simple to administer without use of a computer | Closed response format limits participants to rating pre-determined interpretations, none of which may reflect their own interpretation |
| - Rate likelihood of possible explanations | - Participants rate the likelihood of a number of presented interpretations [written response] | Simple to administer without use of a computer | Closed response format limits participants to rating pre-determined interpretations, none of which may reflect their own interpretation |
| - Imagine the scenario and rate its pleasantness | - Participants imagine the resolution of ambiguity and rate its level of pleasantness [written response] | Less influenced by demand characteristics as participants do not explicitly state their interpretation | Remains unknown exactly how participants interpreted the ambiguous scenario |
| - Imagine sentence and rate level of concern | - Participants indicate whether the event would cause them concern or not [written response] | Less influenced by demand characteristics as participants do not explicitly state their interpretation | Remains unknown exactly how participants interpreted the ambiguous scenario |
| - Alternative attributions measure | - Participants are encouraged to think of as many alternative interpretations to their initial interpretation, which are rated as positive or negative [written response] | Simple to administer without use of a computer | Although alternative interpretations are provided, responses may still be subject to demand characteristics |
| Ambiguous situations task | Following the presentation of an ambiguous scenario, participants answer a series of open-ended (e.g., What would you think in the situation?) and closed questions (e.g., Likert scale to rate the valence of the situation). This paradigm is similar to the ambiguous scenarios test but has been classified separately in this review for clarity [verbal and written response] | Simple to administer without use of a computer. Scenarios may be presented aurally so comprehension is not likely to be influenced by reading ability | The use of multiple response formats may be tiring and/or confusing for participants, especially children. Responses to open-ended questions may be subject to demand characteristics and influence responses to subsequent closed questions |
| Scrambled sentence task | Participants are presented with a series of words in random order that can be unscrambled and combined to form a negative or positive sentence [keyboard response] | Relatively simple to design and administer when comparing positive/negative interpretations. Flexibility for developing a large number of sentences. Demand characteristics can be reduced via used of cognitive load requirement | Potentially more difficult to develop appropriate stimuli for exploring anything other than positive/negative sentences |
| Word sentence association paradigm—traditional version | Participants are presented with a single word (benign or threatening) followed by an ambiguous sentence, and indicate whether the two are related or not. Response time may also be collected [keyboard or mouse response] | Flexibility in exploring the impact of different categories (e.g., negative, positive, disorder-specific) of primes and target sentences. Both self-report and response time data can be collected providing direct and indirect measures of bias | Unless comprehension questions included, cannot be certain participants read both sentence and words; they may respond to one or the other with no comparison made between the two |
| Word sentence association paradigm—modified version | Participants are presented with an ambiguous sentence followed by a single word (benign or threatening), and indicate whether the two are related or not. Response time may also be collected [keyboard or mouse response] | Flexibility in exploring the impact of different categories (e.g., negative, positive, disorder-specific) of primes and target sentences. Both self-report and response time data can be collected providing direct and indirect measures of bias | Unless comprehension questions included, cannot be certain participants read both sentence and words; they may respond to one or the other with no comparison made between the two |
| Lexical decision task - Grammatically correct |
Participants read narrative texts with ambiguous sentences in which certain target words could be either threatening or neutral, and must indicate whether the target word is grammatically correct. Both accuracy and response times are recorded [keyboard or response box] | May measure relatively spontaneous and automatized interpretations (i.e., on-line interpretations at the time of reading) | Text readability may need consideration. Reading ability of participants may influence results, especially for more complex scenarios |
| - A real English word | Participants read narrative texts with ambiguous sentences in which certain target words could be either threatening or neutral, and must indicate whether the target word is a real English word. Both accuracy and response times are recorded [keyboard or response box] | May measure relatively spontaneous and automatized interpretations (i.e., on-line interpretations at the time of reading). Does not necessitate backwards checking of preceding text to provide answers | Text readability may need consideration. Reading ability of participants may influence results, especially for more complex scenarios |
| Reduced evidence of danger task | Children are presented with ambiguous and non-ambiguous spoken stories, and are instructed to determine as fast as possible whether the story was scary or not (i.e., have a good or bad ending). Specifically, children are asked after each sentence whether the story is scary. Children also indicate how they think each story will end, and how they would feel and think if in the situation [verbal response] | Individual differences in reading and writing ability do not affect participant responses; appropriate for children | May be subject to demand characteristics if children are informed that some of the presented stories are threatening |
| Sentence completion task - Complete with single word |
Ambiguous sentences are presented with the final word missing. Participants complete the sentence with as many single-word responses that they can think of, and endorse the one response they believe best completes the sentence [written response] | Allows for testing of whether first interpretation provided is the one most endorsed | Unknown whether multiple responses occur prior to resolution of ambiguity or afterwards, the latter of which may reflect elaboration of responses based on demand characteristics of the task |
| - Complete with as many words as necessary | Incomplete ambiguous scenarios are presented, with participants providing a suitable ending in as many words as they like [written response] | Simple to administer without use of a computer | Responses may be subject to demand characteristics |
| Text comprehension task | Sentence pairs are presented, the first of which is ambiguous with both threatening and non-threatening interpretations. A continuation sentence is then presented corresponding to one of the two possible interpretations. Participants indicate whether the two sentences are related, with response times providing an indication of resolution of ambiguity [keyboard or response box] | Minimisation of demand effects as the critical measure is assessed covertly | Cannot be certain participants read each and every sentence, unless comprehension questions are added |
| Recognition task - Indicate whether recognize |
Participants are presented with ambiguous sentences. In a subsequent recognition task, participants indicate whether or not they recognize disambiguated sentences [keyboard response] | Flexibility in exploring the impact of different scenario categories (e.g., pain-related, social threat) | Interpretation bias assessed retrospectively, so results may reflect negative recall bias |
| - Indicate similarity | Participants are presented with ambiguous sentences. In a subsequent recognition task, rate disambiguated sentences in terms of how similar they are to the original sentence in their meaning [written response] | Flexibility in exploring the impact of different scenario categories (e.g., pain-related, social threat) | Interpretation bias assessed retrospectively, so results may reflect negative recall bias |
| Reading time paradigm | An incomplete ambiguous sentence is presented with either threatening or non-threatening consequences. A disambiguating sentence is then shown which provides either a threatening or non-threatening conclusion to the scenario. Words are presented using a moving window technique, which are displayed when the participant presses a key. Reading time for target words are indicative of interpretation (faster for words congruent with interpretation of the ambiguous sentence) [keyboard reponse] | Allows for exploration of time-course of interpretation bias. Indirect measure of bias less influenced by demand characteristics | No guarantee participants fully read or engage with each sentence, unless comprehension questions are asked |
| Naming task | An incomplete ambiguous sentence is presented with either threatening or non-threatening consequences. A disambiguating sentence is then shown which provides either a threatening or non-threatening conclusion to the scenario. Words are presented one at a time. Naming time (i.e., verbal pronunciation) for target words are indicative of interpretation (faster for words congruent with interpretation of the ambiguous sentence) [verbal response] | Strategic vs. automatic biases may be explored by varying the interval between pre-target and target word. Indirect measure of bias less influenced by demand characteristics | No guarantee participants fully read or engage with each sentence, unless comprehension questions are asked |
| Ambiguous video clips - Participants rate how they would feel |
Participants are shown video clips depicting a stranger approaching the camera and delivering ambiguous lines. Participants rate how they would feel in that situation [written response] | Increased ecological validity | Difficult to explore self-referent biases. Physical characteristics of actors may need to be controlled. Acting quality may influence believability of videos |
| - Participants rate clips | Participants are shown video clips of social interactions with ambiguous endings. Clips are rated in terms of valence, confidence in outcome, perceived escalation of threat, predictability and controllability [written response] | Increased ecological validity | Difficult to explore self-referent biases. Physical characteristics of actors may need to be controlled. Acting quality may influence believability of videos |
| - Participants rate possible explanations on their likelihood | Participants are shown video clips of social interactions with ambiguous endings, and subsequently rate the likelihood of positive, negative, and neutral explanations [written response] | Increased ecological validity | Difficult to explore self-referent biases. Physical characteristics of actors may need to be controlled. Acting quality may influence believability of videos. Closed response format limits participants to rating pre-determined interpretations, none of which may reflect their own interpretation |
| Confederate performing ambiguous behaviors | Participants perform a short speech, during which a confederate performs ambiguous behaviors (e.g., scratching head). Participants subsequently rate confederate behaviors using closed—and open-ended questions [written response] | Increased ecological validity, with participants directly exposed to ambiguous behaviors | Despite attempts at standardization, confederate behaviors may vary between participants. Confederate characteristics may also determine how noticeable ambiguous behaviors are |
| Story-stem paradigm | Children are presented with the beginning of an ambiguous story, and are then asked to complete the story using dolls and other props [behaviors are videoed and later systematically coded] | Suitable for children. Responses do not depend on reading ability | Characteristics of the experimenter and testing environment may influence child behaviors |