trait anxiety and reasoning under uncertainty

www.elsevier.com/locate/paid

Personality and Individual Differences 43 (2007) 827–838

Trait anxiety and reasoning under uncertainty

Luca Bensi *, Fiorella Giusberti

Department of Psychology, University of Bologna, Viale Berti Pichat 5, 40127 Bologna, Italy

Received 1 June 2006; received in revised form 20 December 2006; accepted 2 February 2007Available online 26 March 2007

Abstract

We explored the relationship between trait anxiety and reasoning under conditions of uncertainty. In ouropinion, high trait anxious individuals (HTA) could have an implicit goal of uncertainty reduction. Accord-ing to this, we predicted that in reasoning tasks HTA individuals would try to shorten the length of uncer-tain states by gathering less evidence prior to deciding, compared to low trait anxious ones (LTA). In Study1 we employed several probabilistic reasoning tasks to examine the amount of data requested before mak-ing a decision. In Study 2, we investigated how evidence is weighed up and how hypotheses are tested.Results confirmed our prediction: HTA individuals gathered fewer pieces of evidence, thus jumping to con-clusions, compared to LTA individuals. Groups did not differ in their reasoning if evidence was at theirdisposal from the start. Furthermore, the HTA group jumped to conclusions and provided incorrectanswers in the hypothesis-testing task. We suggest that HTA individuals are particularly concerned withreducing uncertainty, even at the expense of correctness.� 2007 Elsevier Ltd. All rights reserved.

Keywords: Trait anxiety; Reasoning; Uncertainty; Information-gathering; Jump to conclusions style

1. Introduction

Anxiety can be divided into two categories: state anxiety and trait anxiety (Spielberger, 1972).State anxiety is defined as a person’s current level of anxiety, which can be modulated by situa-

0191-8869/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.paid.2007.02.007

* Corresponding author. Tel.: +39 051 2091835; fax: +39 051 243086.E-mail address: [email protected] (L. Bensi).

mailto:[email protected]

828 L. Bensi, F. Giusberti / Personality and Individual Differences 43 (2007) 827–838

tional factors and consists of feelings of tension and apprehension. On the other hand, trait anx-iety (TA) is an individual’s general disposition to become anxious, that is, an anxiety proneness.

According to cognitive models of anxiety (e.g., Eysenck, 1997), TA is associated with a numberof cognitive biases regarding threat-related information, which lead to an exaggeration of thethreat level of external and internal stimuli. TA has been reliably associated with a selective atten-tional bias and with an interpretive bias for threat-related information. To date, there appears tobe good evidence that when presented with both threat-relevant and threat-irrelevant stimuli, highTA (HTA) individuals, compared to low TA (LTA) ones, show increased attention towardsthreat-relevant information (e.g., Mathews & MacLeod, 1994). Furthermore, several studies havefound that when presented with ambiguous stimuli, HTA people make more threatening interpre-tations than LTA people (e.g., Byrne & Eysenck, 1993).

Overall, these results seem to point out that TA is associated with attention towards threat cuesand with the likelihood of perceiving the threatening meaning of ambiguous events. HTA individ-uals are prone to locating threat and believe the world to be more dangerous compared to LTA:they watch out for danger and jump to the most threatening conclusion (Mathews & Mackintosh,1998).

1.1. The psychology of thinking

On the other hand, while research has widely studied the influence of TA on selective informa-tion-processing, the relationship between TA, information-gathering and decision-making, in gen-eral, reasoning, has been explored to a lesser extent.

How we collect and/or use information in order to draw inferences, make decisions or testhypotheses is studied by the psychology of thinking. A traditional division in the fields of reason-ing research is that between deductive and inductive thinking. Deduction involves arriving at con-clusions on the basis of statements, called premises, the truth value of which can be assumed.Induction reasoning is used to arrive at a conclusion on the basis of some evidence: it increasesinformation, but an induction conclusion cannot be guaranteed to be true. This is because a con-clusion may be based on irrelevant evidence, relevant evidence may be ignored, new evidence mayforce one to change one’s mind, or there may be bias in the way evidence is treated (Manktelow,1999).

1.2. Trait anxiety and reasoning under uncertainty

To date, only a few empirical studies have explored if and how TA affects general reasoningprocesses, and knowledge about the thinking of anxiety proneness individuals is sparse. For exam-ple, some studies have found that HTA individuals, compared to LTA ones, display significantlylonger solution times and lower accuracy scores for a series of different inferential reasoning tasks(e.g., Chiappelli & Giusberti, 2001; Mayer, 1977).

An interesting study of Leon and Revelle (1985) analysed thoroughly state/trait anxiety andanalogical reasoning, that is, the process of finding systematic correspondences between a noveltarget situation and a more familiar source situation, thus using knowledge of the source to deriveinferences about the target. These authors studied analogical reasoning under either relaxed (non-time-stressed/reassurance) or stressed (time-stressed/ego-threatened) conditions. The findings

L. Bensi, F. Giusberti / Personality and Individual Differences 43 (2007) 827–838 829

about the relaxed condition showed that the HTA group was both slower and less accurate com-pared to the LTA one. On the other hand, in the stressed condition HTA subjects were faster andless accurate compared to LTA ones.

In any case, we know little regarding how HTA individuals behave when facing uncertain con-ditions and when they have to actively gather and use information in order to make decisions ortest hypotheses. These circumstances are particularly interesting with respect to TA. As a matterof fact, when confronted with uncertainty and possibilities, employing reasoning and inferentialprocesses is a way to increase knowledge, thus obtaining evidence to reach a conclusion (e.g., Hol-land, Holyoak, Nisbett, & Thagard, 1986). In such terms, uncertainty is surely an important topicfor anxiety proneness individuals. Anxiety is generally experienced in response to situations wherethe person is uncertain about an impending outcome (especially of a personally relevant event)(Rachman, 1998). Thus we suggest that people prone to anxiety are strongly concerned withuncertainty.

Following this line of thought, the main purpose of the present study was to explore therelationship between an enduring, stable aspect of personality, that is trait anxiety, andreasoning. Specifically, we looked at reasoning under uncertainty in HTA and LTAindividuals.

2. Study 1

A possible approach to the study of how people behave and decide in uncertain situations isprobabilistic reasoning. Undoubtedly, probability and uncertainty are closely linked. Probabilisticreasoning refers to thought processing in uncertain conditions and concerns the use of evidence indeciding whether a hypothesis is true or not. In a classic probabilistic task, beginning from theprior probabilities of one or more events occurring, pieces of data are searched and/or evaluatedin order to choose which hypothesis is true (e.g., Phillips & Edwards, 1966). Of course, the con-clusion, although plausible, is not guaranteed to be true.

We suppose, as cited early, that anxious people are preoccupied with uncertainty regardingsome events or states because they could imply a threat. This is often a subjective/psychologicalthreat, experienced in social, novel or ambiguous situations. According to this, individualsprone to anxiety should perceive the state of being uncertain as stressful and upsetting just be-cause they are likely to experience anxiety and perceive threat in uncertain circumstances. Insuch terms, they should consider uncertain situations unpleasant and uncomfortable, even ifthey are not facing objective threatening situations or impending danger. In other words, whatcould be unacceptable for HTA individuals is the uncertainty in itself, that is, the mere notknowing what will happen (i.e. whether or not a certain threat will come true), apart from thepositivity or negativity of the outcome (Miceli & Castelfranchi, 2005). As a result, HTA indi-viduals could have an implicit goal of uncertainty reduction, thus lowering anxiety anddiscomfort.

In order to study reasoning and TA, we decided to employ probabilistic inferential tasks unre-lated to anxiety themes, that is, content-neutral, for two main reasons. Firstly, we were concernedwith cold reasoning: we wanted to investigate general reasoning processes in TA, not specific toanxiety-related circumstances (hot reasoning). Moreover, several studies have indeed shown


how individual differences in TA can bias subjective judgments regarding risk scenarios. HTAindividuals are particularly prone to believing that they are at risk for certain negative or harmfulevents (e.g., Kverno, 2000), and they overestimate the probability of negative events occurring(e.g., Butler & Mathews, 1987) compared to LTA individuals. In these terms, anxiety-relatedmaterials could interfere with performances on probabilistic reasoning tasks which involvethreat/danger probability estimates. We chose to use neutral content materials with the aim ofavoiding such interference.

According to our assumptions, we predicted that HTA individuals would act in a way to reduce(if possible) uncertainty present in our reasoning tasks; specifically, we hypothesized that theywould try to shorten the length of the uncertain state. With this in mind, we employed two prob-abilistic reasoning tasks in order to study the amount of data gathered to make a decision in aformal probabilistic task (Bead Tasks) and in hypothetical real-life situations (Scenario Tasks).We supposed that HTA individuals, compared to LTA ones, would be prone to requesting a low-er number of pieces of evidence in order to decide. Accordingly, HTA individuals would be cog-nitively quicker in forming their decisions, precipitating their conclusions and thus ending the task(and the discomfort caused by uncertainty). In other words, we predicted that HTA people wouldshow a jump to conclusions style, a particular reasoning style studied in these last 20 years: peoplewho exhibit a jump to conclusions style tend to gather less evidence in order to make probabilisticdecisions compared to people who do not exhibit such a style (for a review, see Garety & Free-man, 1999).

2.1. Method

2.1.1. ParticipantsSixty participants (31 males and 29 females) filled out the Italian trait version of the State-

Trait Anxiety Inventory (STAI; Pedrabissi & Santinello, 1989). High and Low Trait Anxiousgroups were formed on the basis of a median split on the STAI trait scale: 23 participantswho scored above 38 (the median score of the sample) were included in the High Trait Anxiousgroup (mean age = 27.18, SD = 3.67; M = 11, F = 12), and 37 participants who scored at orbelow 38 were included in the Low Trait Anxious group (mean age = 25.42, SD = 3.45;M = 15, F = 22). The mean STAI Trait scores were as follows: HTA 49.13 (SD = 5.02),LTA 34.29 (SD = 3.06).

2.1.2. MaterialsTo assess trait anxiety we used the Italian version of the STAI-Y2 form (Pedrabissi & Santi-

nello, 1989). This scale has good internal consistency (a = .85–.90) and test–retest reliability(r = .82–.86).

For the Bead Tasks we employed a modified version of probabilistic tasks outlined by the workof Garety, Hemsley, and Wessely (1991). Materials included three pairs of jars, each containing100 beads, differently coloured and in equal but reversed proportions: Pair 1 (jar A: 85 green/15 orange, jar B: 85 orange/15 green); Pair 2 (jar A: 70 red/30 blue, jar B: 70 blue/30 red); Pair3 (jar A: 60 yellow/40 cyan, jar B: 60 cyan/40 yellow).

For the Scenario Tasks, we created three realistic scenarios on the computer in order to exam-ine the amount of data requested to make a decision in less abstract environments.


2.1.3. ProcedureParticipants were submitted to all the conditions of the two tasks. The order of presentation of

Bead Tasks and Scenario Tasks were counterbalanced; the three Bead Task conditions (i.e. jarpairs) and the three Scenario Task conditions (i.e. scenarios) were counterbalanced in each ofthe two tasks. STAI administering was also counterbalanced, that is, participants filled out theinventory either before or after the two tasks.

2.1.3.1. Bead Tasks. The experimenter selected one of the three pairs of jars (e.g., Pair 1), chose oneof the two jars (e.g., jar A, with B taken away) randomly and out of sight, and began to draw out onebead at a time; after each extraction, the experimenter replaced the bead in the jar. Subjects were in-formed about the three pairs of jars and their respective proportions of beads; moreover, they weretold that a priori estimate (probability of jar A or B to be chosen) was equivalent (50:50). Participantshad to decide from which jar of the chosen pair the experimenter was drawing the beads (jar A or jarB). They could ask to see as many beads as they wished before providing the answer (thus ending thetask). In this way, it was possible to investigate the amount of data (drawings) required by partici-pants in order to make the decision. This procedure was the same for the other two pairs of jars.The drawings were prearranged: beads were taken out following a fixed sequence. In this way, eachsubject could decide how much information was necessary but, data being equal, he/she provided ajudgment on the same evidence received by all the others. The three planned series were:

Pair 1 (85:15) = GGGOGGGGOGGGGGGGGOGPair 2 (70:30) = RRBRRBRRBRRRBRRBRRBRPair 3 (60:40) = YYCYCYYCYCYYCYCYYCYCG = green; O = orange; R = red; B = blue; Y = yellow; C = cyan

2.1.3.2. Scenario Tasks. Three realistic scenarios were presented on a computer screen. Each con-dition was preceded by the respective instructions. Participants could request all the evidence theyneeded prior to deciding by pressing the ‘‘SPACE-BAR’’ on the computer keyboard. The follow-ing is an example scenario:

2.1.4. Scenario 3‘‘Imagine being at your home and watching a quiz. The host alternately asks two competitors

general knowledge questions. You are trying to understand which competitor is better prepared.Pressing the ‘‘SPACE-BAR’’ key, you will be able to see, for each question, if the competitorsanswered correctly or not. When you believe you know who is the best, provide me with youranswer’’.

Sequences of information were fixed for each of three conditions, and thus each participant waspresented, evidence being equal, with the same pieces of information.

2.2. Results

Regarding Bead Tasks, we performed a 2 (anxiety: HTA vs. LTA) · 3 (bead ratio: 85:15 vs.70:30 vs. 60:40) mixed ANOVA to study the influence of anxiety and different bead ratio condi-


tions on participants’ performances. Results showed a significant main effect of anxiety, F(1,58) = 17.57, p < .01, and bead ratio, F(2, 116) = 62.53, p < .01, but no interaction between anxietyand bead ratio, F(2, 116) = .02. The HTA group requested significantly less evidence in order tomake a decision than the LTA group, in each of the three conditions: 85:15 (t(58) = 3.56, p < .01),70:30 (t(58) = 2.78, p < .01), 60:40 (t(58) = 2.50, p < .05). Moreover, both groups were sensitive toconditions, that is, they gathered evidence accordingly to bead ratio: LTA = 85:15 vs. 70:30,t(36) = 4.03, p < .01; 85:15 vs. 60:40, t(36) = 9.08, p < .01; 70:30 vs. 60:40, t(36) = 4.86, p < .01;HTA = 85:15 vs. 70:30, t(22) = 5.17, p < .01; 85:15 vs. 60:40, t(36) = 6.36, p < .01; 70:30 vs.60:40, t(22) = 3.54, p < .01. Mean scores across the three conditions are reported in detail inFig. 1.

Regarding Scenario Tasks, we performed a 2 (anxiety: HTA vs. LTA) · 3 (scenario: Scenario 1vs. Scenario 2 vs. Scenario 3) mixed ANOVA to study the influence of anxiety and different beadratio conditions on participants’ performances. Results showed a significant main effect of anxi-ety, F(1, 58) = 22.92, p < .01, and scenario, F(2, 116) = 18.05, p < .01, but no interaction betweenanxiety and scenario, F(2, 116) = .55. The HTA group gathered a lesser amount of data beforeproviding an answer compared to the LTA group, in each of the three scenarios (Scenario 1:t(58) = 4.11, p < .01; Scenario 2: t(58) = 4.35, p < .01; Scenario 3: t(58) = 3.40, p < .01). Meanscores across the three scenarios are reported in detail in Fig. 2.

2.3. Discussion

The results of this study support our assumption: in Bead Tasks, HTA individuals gatheredfewer drawings in order to conclude that the experimenter had chosen a certain jar rather thanthe other, that is, they jumped to conclusions. In other words, the HTA group were quicker inchoosing a jar than the LTA group. This effect was also found in every condition of the ScenarioTasks, that is, in more real-life and less abstract decision-making environments.

7,73

5,86

4,34

9,16

7,27

5,64

0

2

4

6

8

10

12

85-15 pair 70-30 pair 60-40 pair

num

ber

of d

raw

ings

to

deci

sion

HTA LTA

Fig. 1. Bead Tasks – Number of drawings requested prior to decision-making.

12,5211,26

13,8615,4

14,4316,27

0

2

4

6

8

10

12

14

16

18

20

Scenario 1 Scenario 2 Scenario 3

amou

nt o

f ev

iden

ce t

o de

cisi

on

HTA LTA

Fig. 2. Scenario Tasks – Amount of evidence requested prior to decision-making across the three scenarios.


Overall, such findings are consistent with our hypothesis. The proneness to make a hasty deci-sion showed by the HTA group in Bead Tasks and Scenario Tasks could serve the same purpose:finding a conclusion, thus not remaining in an uncertain state. According to this, reducing uncer-tainty could be an implicit goal of high trait anxious individuals.

3. Study 2

Findings of Study 1 are consistent with a tendency to request less evidence to make a decision,but they could also reflect a particular approach to probability employed by HTA individuals. Inother words, jumping to conclusions could be the result of a different way of weighing up evidencein order to make a decision in probabilistic tasks and not a cognitive way of reducing uncertainty.For this reason, we carried out a second study to examine how HTA individuals perform on prob-ability tasks when: (1) they have at their disposal all the information from the beginning (CoinTask 1); (2) they are free to ask for the amount of evidence they need (Coin Task 2). We predicted,according to our general assumption, that HTA and LTA participants would perform similarly inCoin Task 1 (in this case, uncertainty-reduction, that is, jumping to conclusions, is not possible),but differently in Coin Task 2 (uncertainty-reduction is possible: HTA individuals would requestless data and jump to conclusions).

Furthermore, the first study did not provide any hints with respect to the possibility that col-lecting less evidence could impair decisions and affect the number of correct answers from HTApeople. We suggest that HTA individuals are mainly preoccupied with uncertainty and withreducing uncertain states, even if at the expense of correctness. According to this, we predictedthat HTA participants, jumping to conclusions, would gather an insufficient amount of infor-mation and, for this reason, they would perform worse compared to LTA participants (CardTask).


3.1. Method

3.1.1. ParticipantsFifty-one (23 males and 28 females) participants filled out the Italian trait version of the STAI.

High and Low Trait Anxious groups were formed on the basis of a median split on the STAI traitscale: 24 participants who scored above 39 (the median of the sample) were included in the HighTrait Anxious group (mean age = 24.89, SD = 4.31; M = 13, F = 11), and 27 participants whoscored at or below 39 were included in the Low Trait Anxious group (mean age = 26.80,SD = 5.17; M = 14, F = 13). The mean STAI Trait scores were as follows: HTA 48.04(SD = 4.90), LTA 33.44 (SD = 3.59).

3.1.2. ProcedureThe three tasks (Coin Task 1, Coin Task 2, Card Task) were submitted to all participants on a

computer screen. Trials were carried out individually and the order of presentation of the tasksand STAI administering were counterbalanced.

3.1.3. Coin Task 1Subjects were presented with a series of coin tosses and had to decide if, in their opinion, the

coin could be a trick one, thus favouring a certain outcome, or if it was genuine. ‘‘HEADS’’ ratioacross conditions was always 3:4. The five conditions were: 4 tosses/3 ‘‘HEADS’’; 12 tosses/9‘‘HEADS’’; 20 tosses/15 ‘‘HEADS’’; 40 tosses/30 ‘‘HEADS’’; 60 tosses/45 ‘‘HEADS’’.

3.1.4. Coin Task 2As in the previous task, participants had to decide if a coin was a trick one or genuine. Pressing

‘‘SPACE-BAR’’ on the keyboard, the outcome of a coin toss appeared, one at a time, on the com-puter screen. ‘‘HEADS’’ ratio was 3:4. Participants could provide their answer to the experi-menter whenever they wished.

3.1.5. Card TaskWe used a modified version of a Card Task employed in previous studies in order to explore

hypothesis testing skills (e.g., Young & Bentall, 1995). In this task, participants were presentedwith a sequence of card pairs, one pair at a time, and were asked to choose one among eight char-acteristics (X or T letter, large or small letter, black or white letter, left or right card) in order tofind the correct one (i.e. the characteristic selected by the experimenter). Participants chose a char-acteristic and they selected at each presentation the card (left or right) according to that charac-teristic (i.e. for X letter, they had to point to the card that always contained the X letter,disregarding the other characteristics).

No ‘‘correct’’ characteristic was fixed by the experimenter. After a planned number of cardpairs, the experimenter provided positive ‘‘feedback’’ information, always saying ‘‘correct’’ as re-gards the card (left or right) chosen. The card pair sequence was built in such a way that it was notpossible to be mathematically sure about the correctness of the answer before reaching the fifthpiece of feedback information: as a matter of fact, only in this case all the alternative hypothesescould be discarded (see card sequence in Fig. 3). Therefore, the experimenter could observe theamount of feedback evidence requested by participants before reaching a decision.

3 random non-feedback card pairsFirst feedback card pair

X3 random non-feedback card pairsSecond feedback card pair

T

3 random non-feedback card pairsThird feedback card pair

X3 random non-feedback card pairs

Fourth feedback card pair

T

3 random non-feedback card pairsFifth feedback card pair

X

Fig. 3. Card Task – Feedback cards – Whatever the chosen characteristic was, it was possible to be sure about thecorrectness of the answer only by reaching the fifth piece of feedback information.


3.2. Results

No significant differences were found between the two groups on Coin Task 1: both HTA andLTA groups judged the coins alike (trick/genuine) in each of the five conditions (4:3 – all partic-ipants answered ‘‘genuine coin’’; 12:9, v2(1) = 0.03, p = .85; 20:15, v2(1) = .98, p = .32; 40:30,v2(1) = .09, p = .75; 60:40, v2(1) = .09, p = .75).

In the Coin Task 2, the HTA group needed a significantly lower number of tosses in order tomake a decision (M = 16.04, SD = 5.08) than the LTA group (M = 19.88, SD = 4.31)(t(49) = 2.92, p < .01).

In the Card Task, the HTA group decided after a significantly lower number of feedback cards(M = 3.87, SD = 1.07) compared to the LTA group (M = 4.85, SD = 1.02) (t(49) = 3.31, p < .01).The HTA group provided significantly fewer correct answers (33.33%) than the LTA group(70.37%) (v2(1) = 6.99, p < .01). Obviously, we considered any decision made before the fifth feed-back card as wrong, though it was consistent with the data collected until then. Such an answer,indeed, could not be due to a correct assessment, checking and discarding of all alternativehypotheses.


3.3. Discussion

The results of this second study support our hypotheses. Groups did not differ in their behav-iour if they had to decide on the likelihood of an event and all the information was at their dis-posal from the start (Coin Task 1). On the other hand, HTA and LTA participants performeddifferently when they were asked to actively gather information and then make a decision (CoinTask 2). In accordance with our general hypothesis, when HTA individuals could choose howmany instances of evidence they needed to provide an answer, they jumped to conclusions, thusending the uncertain state.

Furthermore, the HTA group gathered insufficient information in the hypothesis-testing taskand this behaviour affected the correctness of their answers (Card Task). In general, they decidedafter a number of feedback cards that could not ensure that their conclusion was the right one.According to our hypothesis, HTA individuals jumped to conclusions thus ending uncertainty,gathering insufficient information: therefore, they provided incorrect answers.

4. General discussion

The main aim of the present studies was to investigate the relationship between a stable char-acteristic of personality, trait anxiety, and reasoning processes in the event of facing uncertainty.We supposed that HTA individuals would find uncertain situations particularly unpleasant anduncomfortable. According to this, our general hypothesis was that HTA individuals would jumpto conclusions in our reasoning tasks because they have an implicit goal of reducing uncertainstates, thus eliminating discomfort.

Overall findings support such a hypothesis: HTA individuals, compared to LTA individuals,gathered less information and made a quicker decision under uncertain conditions. We found sucha reasoning pattern in content-neutral tasks, not related to anxiety themes, concerning both formaland hypothetical real-life decisions: this suggests that the tendency to request less information toreach a conclusion in probabilistic and uncertain conditions is a general style of HTA individuals.

Differences between HTA and LTA individuals were found in reasoning tasks in which infor-mation has to be actively collected in order to reach a conclusion. When all the evidence requiredin order to make a decision was available from the beginning and no further information had tobe requested, both groups performed in the same manner. According to this, HTA individualsshow a cognitive reasoning style that affects performance only when evidence has to be gatheredto reach a conclusion.

It is notable that these effects do not appear to be the result of a disregard for task demands. In fact,in Bead Task 1 (Study 1) they took account of the three different base chances and provided theiranswer accordingly, that is, they gathered less evidence in the 85:15 condition and more evidencein the 60:40 condition. We can suppose that the HTA group would have decided without a consid-eration of base likelihoods if their goal had only been to end the task. In such terms, HTA individualsjumped to conclusions but they were not only in a hurry to conclude: they paid attention to the var-ious evidence provided in order to make a decision. Therefore, we suggest that the objective of HTAindividuals is to reduce anxiety providing, at the same time, a suitable conclusion. According to this,in the Card Task (Study 2) HTA individuals reduced the length of uncertainty, gathering a smaller


amount of information but, at the same time, they provided a plausible answer because the evidencewas in agreement with their hypothesis. HTA individuals jumped to conclusions and, for this reason,they were mostly incorrect: in any case, they reduced uncertainty and reached an appropriate thoughincorrect decision on the basis of data collected up until the moment of decision-making. It should beunderlined that a reasonable conclusion could be reached in all our reasoning tasks. These presented,in fact, uncertain but not ambiguous situations: evidence, indeed, always supported a hypothesisrather than another. Ambiguous situations, on the other hand, could lead to various interpretationsso that the conclusion may be dubious and questionable. For instance, in the condition 85:15 of theBead Tasks (Study 1), if the outcomes of the drawings had been one green–one orange, repeatedly,reaching a sensible decision would have been very difficult.

It is interesting to compare our results with Leon and Revelle’s (1985) findings. In the presentstudy, the HTA group were faster (in terms of the amount of data gathered) compared to the LTAgroup in all conditions. On the other hand, Leon and Revelle showed how HTA individuals wereslower (in terms of response) on their analogical reasoning performances compared to LTA onesin a relaxed condition, and conversely, how they were faster in a stressed condition. Even thoughthe reasoning tasks of the two studies are not conceptually similar (reasoning under uncertaintyvs. analogical reasoning) nor is the dependent variable under analysis (amount of data gatheredvs. response time), taken together these results are quite remarkable. As a matter of fact, it ap-pears that HTA individuals do not show a unique pattern of behaviour on reasoning tasks, in-stead, they perform faster or slower depending on several factors, like kind of reasoninginvolved, condition request and emotional state. Anyway, we can assume that if our hypothesisis well-grounded, the aim to reduce uncertainty could be accomplished in different ways accordingto the characteristics of reasoning tasks, thus eliciting different strategies: focusing on accuracywhen all data are available from start, or speeding up decision when pieces of evidence are freeto be gathered or under stressful conditions.

To summarise, it appears that HTA individuals, in general, were more concerned with eliminat-ing discomfort and providing an ‘‘acceptable’’ conclusion rather than with solving the task cor-rectly. In other words, the behaviour exhibited by HTA individuals looks like a sort ofheuristic reasoning (Tversky & Kahneman, 1974). When evidence seems to point to a certainhypothesis, HTA individuals could be particularly prone to accepting that hypothesis. Neverthe-less, a swift decision may not be the best one if the goal is correctness. On the other hand, suchheuristic reasoning is likely to be functional for the HTA individuals’ goal: it allows one to reduceuncertainty quickly and (if possible) in a suitable way.

Finally, the present study has not dealt with a contingent, situational anxious state, that is, stateanxiety. Future studies should investigate the influence of state anxiety on reasoning of HTA indi-viduals under uncertain conditions. Furthermore, it should be interesting to extend and comparethe findings of our work by studying probabilistic reasoning and trait anxiety with more salientrelated material, that is, anxiety themes.

References

Butler, G., & Mathews, A. (1987). Anticipatory anxiety and risk perception. Cognitive Therapy and Research, 11(5),551–565.


Byrne, A., & Eysenck, M. W. (1993). Individual differences in positive and negative interpretive biases. Personality and

Individual Differences, 14(6), 849–851.Chiappelli, M., & Giusberti, F. (2001). Delirio e ragionamento. Uno studio sperimentale sul pensiero nei disturbi

mentali. Minerva Psichiatrica, 42(2), 107–117.Eysenck, M. W. (1997). Anxiety and cognition: a unified theory. London: Erlbaum.Garety, P. A., & Freeman, D. (1999). Cognitive approaches to delusions: a critical review of theories and evidence.

British Journal of Clinical Psychology, 38(2), 113–154.Garety, P. A., Hemsley, D. R., & Wessely, S. (1991). Reasoning in deluded schizophrenic and paranoid patients: Biases

in performance on a probabilistic inference task. Journal of Nervous and Mental Disease, 179(4), 194–201.Holland, J., Holyoak, K. J., Nisbett, R. E., & Thagard, P. (1986). Induction. Cambridge: Brandford Books/MIT Press.Kverno, K. S. (2000). Trait anxiety influences on judgments of frequency and recall. Personality and Individual

Differences, 29(3), 395–404.Leon, M. R., & Revelle, W. (1985). The effects of anxiety on analogical reasoning: a test of three theoretical models.

Journal of Personality and Social Psychology, 49(5), 1302–1315.Manktelow, K. (1999). Reasoning and thinking. Hove: Psychology Press.Mathews, A., & Mackintosh, B. (1998). A cognitive model of selective processing in anxiety. Cognitive Therapy and

Research, 22(6), 539–560.Mathews, A., & MacLeod, C. (1994). Cognitive approaches to emotion and emotional disorders. Annual Review of

Psychology, 45, 25–50.Mayer, R. E. (1977). Problem-solving performance with task overload: Effects of self-pacing and trait anxiety. Bulletin

of the Psychonomic Society, 9(4), 283–286.Miceli, M., & Castelfranchi, C. (2005). Anxiety as an ‘‘epistemic’’ emotion: an uncertainty theory of anxiety. Anxiety,

Stress, and Coping, 18(4), 291–319.Pedrabissi, L., Santinello, M., 1989. STAI, State-Trait Anxiety Inventory, Forma Y: Manuale. Firenze; Organizzazioni

Speciali.Phillips, L. D., & Edwards, W. (1966). Conservatism in a simple probabilistic inference task. Journal of Experimental

Psychology, 72(3), 346–354.Rachman, S. (1998). Anxiety. Hove, UK: Psychology Press.Spielberger, C. D. (1972). Anxiety: Current trends in theory and research: I. Oxford, England: Academic Press.Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: heuristics and biases. Science, 185(4157),

1124–1131.Young, H. F., & Bentall, R. P. (1995). Hypothesis testing in patients with persecutory delusions: Comparison with

depressed and normal participants. British Journal of Clinical Psychology, 34(3), 353–369.

trait anxiety and reasoning under uncertainty

Documents