Heuristic Traps In Recreati Onal Avalanche Accidents .
Avalanche News, No. 68, Spring 2004Heuristic Traps in Recreational Avalanche Accidents:Evidence and Implicationsby Ian McCammonAuthors note: This article extends the findings I presented at the 2002 ISSW in Penticton, BC. A full version of thearticle, including a description of the statistical methods used, appeared in the Avalanche Review (Vol. 22, nos. 2& 3). You can download the two-part article at www.snowpit.com.highmark a slope they believe is safe. And then theytrigger an avalanche that buries one or more of them.In hindsight, the danger was often obvious beforethese accidents happened, and so people struggle toexplain how intelligent people with avalanche trainingcould have seen the hazard, looked straight at it, andbehaved as if it wasn’t there.Several years ago, my buddy Steve died in anavalanche. It was a stormy day and the avalanchedanger was high, but Steve and his partners felt that bychoosing a familiar route and carefully managing theirexposure, they could stay out of trouble. After all, theywere experienced backcountry skiers with avalanchetraining. Steve, the most skilled of the group, had justvisited the area less than a week before.Heuristic traps in avalanche accidentsTwo hours into their tour, they met another party ofskiers headed for the same pass and the low-angleslopes on the far side. They briefly discussed theavalanche conditions, and agreed that prudent routeselection was the key to safety that day. But tenminutes later, as Steve’s group broke trail across ashallow, treed slope, they triggered an avalanche thatswept down on them from above. The avalanchecaught all three skiers, seriously injuring one man andcompletely burying Steve. The other party witnessedthe accident and came to the rescue, but by the timethey dug Steve out, he was dead.So how do people come to believe that a slope is safe,even when they are faced with likely evidence that itisn’t? One possible explanation is that people aremisled by unconscious heuristics, or rules of thumb,that guide most of our decisions in everyday life. 1Such heuristics work well for dealing with routinerisks such as driving, using crosswalks, or avoidingsocial embarrassment. But as we’ll see, avalanchespresent a unique hazard that renders some of ourheuristics irrelevant, and in some cases dangerouslymisleading. When a rule of thumb gives us a grosslyinaccurate perception of a hazard, we fall into what isknown as a heuristic trap.In the aftermath of the accident, some people felt thatSteve died because he took foolish risks that day.Traveling in avalanche terrain during high hazard, theysaid, was reckless. They believed Steve’s group hadignored obvious signs of danger, and that they weretempting fate by crossing under an avalanche path insuch conditions. The explanation sounded reasonable.Six heuristics in particular are recognized as beingwidely used in our daily decisions: familiarity,consistency, acceptance, the expert halo, socialfacilitation and scarcity. 2 Because these heuristicswork so well and because we’ve used them for muchof our lives, we are largely unaware of using them,even when we are making critical decisions. Suchconditions are fertile ground for heuristic traps.But it didn’t match what I knew about Steve. Weeksearlier, I had shared a lift ride with him at a local skihill, and we had reminisced about our climbingadventures years before. We laughed about how Steveused to love leading thin, difficult routes, often highabove his protection. But things were different now, hesaid. He told me about his wife and his beautiful fouryear-old daughter, and how his days of being recklesswere over, and how the time for raising his family hadbegun. He still loved to ski and climb, he said, but nowit was more about enjoying the outdoors and cominghome afterwards than about taking risks. When hedied, it was on a popular route in familiar terrain, on aslope traversed by dozens of people every season, in aplace that he believed was safe.To study the possible influence of these six heuristictraps in avalanche accidents, I reviewed 715recreational accidents that took place in the UnitedStates between 1972 and 2003. Data for the studycame from records maintained by the ColoradoAvalanche Information Center, published accounts inthe Snowy Torrents (Williams and Armstrong, 1984;Logan and Atkins, 1996), the Westwide AvalancheNetwork, the Cyberspace Snow and Avalanche Center,avalanche forecast center annual reports, and variousInternet and newspaper resources.We will see that there is good evidence that manyavalanche victims fell prey to one or more heuristictraps. But because this study is based on accident data,it can only demonstrate correlations between victims’behavior and the presence of heuristic trap cues.As sad as this accident was, the real tragedy is thatsimilar stories unfold in accident after accident, yearafter year. An experienced party, often with avalanchetraining, makes a crucial decision to descend, cross, or1
Without doing controlled experiments on people’sbehavior in avalanche terrain (which would beproblematic, to say the least), it is not possible toconclusively establish causation of accidents byheuristics traps. Thus, the conclusions of this studyshould be viewed as preliminary – other causativefactors may be at work. Nevertheless, we will see thatexperimental results from other fields of humanbehavior support many of the findings.finding is consistent with the frequently-madeobservation that most avalanche victims appear tohave ignored obvious signs of instability (Fesler, 1980;Smutek, 1980; Jamieson and Geldsetzer, 1996; Atkins,2000; Tremper, 2001). Importantly, there were nocases in the data set where all of the hazard indicatorswere known to be absent.Evaluating decisions by avalanche victimsIf avalanche victims were in fact influenced byheuristic traps, we would expect to see the evidence intheir decisions. Specifically, when trap cues werepresent immediately prior to the accident, susceptiblevictims would be less objective about the avalanchehazard and would tend to expose themselves to morehazard than they would when the trap cues wereabsent. In other words, in accidents where victims fellprey to heuristic traps, the presence of heuristic trapcues would correlate with greater exposure toavalanche hazard.Figure 1. Exposure score frequencies for all accidentsin this study, including those where little informationwas available (N 715).To approximate the objective hazard faced by eachparty prior to the accident, I computed an exposurescore that was a linear combination of seven easilyrecognized indicators of avalanche hazard (Table 1).To minimize reporting biases, I chose indicators thatwould have been readily apparent not only to thevictims, but also to any witnesses, rescue parties orinvestigators.The blatancy of the hazard in avalanche accidentswould be understandable if most victims had littleunderstanding of avalanches. Unfortunately, this doesnot seem to be the case. When accidents parties arecategorized by the training level of the most skilledperson in the party (Table 2), we find that almost halfof the parties contained at least one person (often theleader) who had formal avalanche training and knewnot only how to recognize the hazard, but also how toavoid or mitigate it. Almost two thirds of the partieswere aware of the avalanche hazard, and stillproceeded into the path anyway. Even more telling isthe fact that exposure scores did not significantlydecrease with training.3 Thus, all four levels of trainingappeared potentially susceptible to heuristic traps.The distribution of exposure scores shows that mostvictims proceeded into the avalanche path in the faceof ample evidence of danger (Figure 1). Almost threequarters of all accidents occurred when there werethree or more obvious indicators of the hazard. ThisIndicatorDescriptionObvious pathDistinct start zone, path,runout, trim lines or a knownavalanche path.82%Recent loadingLoading by snowfall 15 cmand/or wind in the last 48hours.66%Terrain trapObvious terrain features suchas cliffs, gullies or dense treesthat increased the severity ofthe slide.58%Posted hazardConsiderable, high or extremehazard posted for the region.55%RecentavalanchesIn the immediate area, withinthe last 48 hours.35%Thaw instabilityAbove-freezing airtemperatures or rain at the timeof the incident.20%Collapsing, cracking, hollowsounds or low stability testscores noted by the victims orthe rescue party.17%Instability o training; displayed noawareness of the avalanchehazard.34%24.3AwareGeneral awareness of theavalanche hazard; took noprecautions prior to theaccident.24%30.1BasicFormal avalanche training;consciously took groupmanagement precautions priorto the accident.28%30.9Adv.Extensive formal training;displayed ongoing avalancheand terrain awareness and riskmanagement. Performedmeaningful snow stability tests.15%33.5Table 2. Training categories used in this study. Frequencydenotes the percentage of accidents where training wasknown or could be reliably inferred (N 484).Table 1. Hazard indicators used in this study.Frequency column denotes the percentage of allaccidents where the indicator was present (N 715).2
A number of investigators have suggested that partysize may have played a role in decisions leading up toavalanche accidents. A “risky shift,” or the tendencyof larger groups to take more risk, has been discussedfrequently in the literature. As shown in Figure 2, thereis a significant variation in exposure score by partysize. It appears that people traveling alone and peopletraveling in parties of six to ten exposed themselves tosignificantly more hazard than people traveling inparties of four and more than ten people.Figure 3. Exposure scores by training in familiar andunfamiliar terrain. Mean values and 95% confidenceintervals are shown. Parties with advanced trainingshowed a notable increase in risky decisions when infamiliar terrain.with the highest level of training (Figure 3), whoexposed themselves to significantly more hazardindicators in familiar terrain. There was a marginallysignificant increase in exposure scores for parties oftwo people.Figure 2. Exposure score variation by accident partysize. Mean values and 95% confidence intervals areshown. Parties of 3–5 people and parties of morethan 10 exposed themselves to the fewest number ofhazard indicators prior to the accident (N 631).Apparently, there is a tendency among highly trainedaccident parties to make riskier decisions in familiarterrain than they do in unfamiliar terrain. Certainly, anintimate knowledge of terrain features, local avalanchehistory, snowpack structure, or the effects of skierstabilization might have contributed to this tendency.But given the large number of accidents that happenedin familiar terrain, it appears that these parties greatlyoverestimated the degree to which familiar slopes weresafer. Remarkably, parties with advanced training thatwere traveling in familiar terrain exposed their partiesto about the same hazards as parties with little or notraining. In some respects, familiarity seems to havenegated some of the benefits of avalanche training.So far, we’ve seen that many avalanche victimsappeared to ignore obvious signs of avalanche danger,regardless of their level of training. We’ve also seenthat party size correlates with different degrees ofexposure to avalanche hazard at the time of theaccident. In the next six sections, we’ll review each ofthe six heuristic traps, and examine how cues for thesetraps correlate with greater exposure by training leveland party size. In other words, we’ll look at how eachtrap may have influenced these victims, and why thesetraps may have been difficult for some parties toavoid. Next, we’ll look at the possible cumulativeeffects of heuristics traps, and which groups were mostsusceptible. Finally, we’ll conclude by examiningwhat all this might mean for avalanche education.Trap #2: ConsistencyOnce we have made an initial decision aboutsomething, subsequent decisions are much easier if wesimply maintain consistency with that first decision.This strategy, known as the consistency heuristic,saves us time because we don’t need to sift through allthe relevant information with each new development.Instead, we just stick to our original assumptions aboutthe situation. 5 Most of the time, the consistencyheuristic is reliable, but it becomes a trap when ourdesire to be consistent overrules critical newinformation about an impending hazard.Trap #1: FamiliarityThe familiarity heuristic relies on our past actions toguide our behavior in familiar settings. Rather than gothrough the trouble of figuring out what is appropriateevery time, we simply behave as we have before inthat setting. 4 Most of the time, the familiarity heuristicis reliable. But when the hazard changes but the settingremains familiar, this rule of thumb can become a trap.To determine if there was evidence of the consistencytrap in avalanche accidents, I compared exposurescores of accident parties that had either high or lowcommitment to entering the path that eventuallyavalanched. Highly committed groups had a statedgoal that they were actively pursuing or a goal theywere motivated to achieve because of approachingdarkness, timing or other constraints (253 cases).Groups with low commitment were not motivated toTo determine if there was evidence of the familiaritytrap in avalanche accidents, I compared exposurescores of accidents that happened in terrain that wasfamiliar (211 cases) or unfamiliar (56 cases) to theaccident party. Taken as a whole, all groups showed asignificant increase in exposure scores in familiarterrain. The effect was most pronounced in parties3
achieve a specific goal; the accident typically occurredduring the course of routine recreational activities (138cases).individuals present in avalanche accident partiesduring the study period, females had a slightly lowerchance of being caught in avalanches then males.Furthermore, as shown in Figure 4, women appearedto avoid participating in parties where they had thehighest probability of being caught.Taken as a whole, exposure scores of all groupsshowed a significant increase when commitment of theparty was high. Among different training levels, theeffect was marginally significant for parties with basicand advanced training. Among different party sizes,the effect was marginally significant for parties ofthree people and significant for parties greater thanfour people. One might argue that any increase inexposure score is simply due to the fact that accidentparties were more likely to commit to skiing orhighmarking a slope when there was new snow, andthus conditions were more hazardous. However, acomparison of avalanche hazard ratings between highcommitment and low-commitment groups showed nocorrelation. 6 Thus, it appears that accident parties whofelt highly committed to enter an avalanche path did infact take more risks than parties who were lesscommitted. This finding is consistent with theobservations of other investigators, most notablyFredston and Fesler (1994) and Tremper (2001).Figure 4. Percentage of females present in accidentparties (columns) and the average percent of each partycaught (line graph). Women appeared to avoid thosegroups where they had the highest chances of beingcaught.Trap #3: AcceptanceThe acceptance heuristic is the tendency to engage inactivities that we think will get us noticed or acceptedby people we like or respect, or by people who wewant to like or respect us. We are socialized to thisheuristic from a very young age, and because we areso vulnerable to it, it’s no surprise that it figuresprominently among the heuristic traps embedded inadvertising messages.The increased exposure of mixed-gender accidentparties may well have been due to reliance on thegender acceptance heuristic by the male partymembers. In other words, males may have been morewilling to expose themselves (and other partymembers) to greater avalanche hazard when there werewomen in the group because such behavior wasviewed by the men as being more likely to gain therespect or acceptance of the women in their party.Certainly, this behavior matches conventional wisdomregarding the conduct of some avalanche victims, asdiscussed by Fredston, Fesler and Tremper (1994) andTremper (2001, p. 226). It is also consistent withrecent findings on the behavior of men in the presenceof women (see, for example, Roney, et al 2003).One of the more familiar forms of this heuristic isgender acceptance, or engaging in activities that webelieve will get us accepted (or at least noticed) by theopposite sex. For men, this heuristic often manifestsitself in certain types of risk-taking behavior,particularly during adolescent and early adult years.Various studies have established that under certaincircumstances, men in the presence of female peerswill behave more competitively, aggressively, orengage in riskier behaviors.Trap #4: The Expert HaloIn many recreational accident parties, there is aninformal leader who, for various reasons, ends upmaking critical decisions for the party. Sometimestheir leadership is based on knowledge and experiencein avalanche terrain; sometimes it is based on simplybeing older, a better rider, or more assertive than othergroup members. Such situations are fertile ground forthe expert halo heuristic, where an overall positiveimpression of the leader within the party leads them toascribe avalanche skills to that person that they maynot have.To see if the gender acceptance heuristic may haveplayed a role in avalanche accidents, I comparedexposure scores from accidents involving mixedgender parties (109 cases) with those of all-maleparties (371 cases). Across all groups, accident partiesthat included women had a significantly higherexposure score. This difference in exposure score didnot vary by group size, but there were notabledifferences by level of training. Parties with awarenessof the avalanche hazard but no formal training (the“aware” training category described in part 1 of thisarticle) showed a significant increase in exposurescores when women were present.To see if there was evidence of the expert haloheuristic in recreational avalanche accidents, Icompared the exposure scores of parties that had aclear, identifiable leader (133 cases) with the exposurescores of parties that had no identifiable leader or theThe increase in the exposure score of accident partiesthat included women does not appear to be a result ofthose women taking more risks. Of the 13554
leadership was unclear (465 cases). Across all groups,parties with an identifiable leader had a significantlyhigher exposure score, but the actual differencesdepended greatly on the level of training of the leader.As shown in Figure 5, the difference in exposure scorewas quite pronounced for those parties who were ledby someone with minimal or no avalanche skills. Whatis surprising about this trend is that untrained partieswith no leader (who presumably made decisions bysome type of consensus process) exposed themselvesto less hazard than they would have if they wererelying on an unskilled leader. In other words,unskilled parties seemed to attribute more avalancheknowledge to their leader than to themselves, evenwhen that leader had no such knowledge.Figure 6. Variation of exposure scores by group sizeand leadership. Decisions by leaders in recreationalaccident parties appeared to get worse as group sizeincreased, compared to the no leader condition.Trap #5: Social FacilitationSocial facilitation is a decisional heuristic where thepresence of other people enhances or attenuates risktaking by a subject, depending on the subject’sconfidence in their risk taking skills.7 In other words,when a person or group is confident in their skills, theywill tend to take more risks using those skills whenother people are pre
The familiarity heuristic relies on our past actions to guide our behavior in familiar settings. Rather than go through the trouble of figuring out what is appropriate every time, we simply behave as we have before in that setting.4 Most of the time, the familiarity heuristic is reliable. But when the hazard changes but the setting
heuristic functions and not all of them are useful during the search, we propose a Topology-based Multi-Heuristic A*, which starts as a normal weighted A* [18] but shifts to Multi-Heuristic A* by adding new heuristic functions to escape from local minima. II. R ELATED W ORK There has been active research on path planning for tethered mobile robots.
means of a thermo drain or pop drain. Also, when multiple traps are installed in a trap station, insulating the traps can provide freeze protection. Thermo Drains are installed in a Tee ahead of 200 Series traps or replace the drain plug directly in the body of specially machined 800 Series traps
types of traps: swim-in bait traps, swim-in bait traps with de- coys, floating bait traps, and decoy traps. Most of the ducks were captured in the spring with Mallard the most numerous species followed by Blue-winged Teal and Wood Duck. Trap- ping mortality averaged 4.5% with mink and raccoon respon- sible for most known causes of death. METHODS
Fish are breed in spawning traps. There are two different types of traps, 1.5-liter transparent traps that can be used to spawn up to 8 fish per trap and 2-liter opaque traps that can hold 12 fish. In H221, clean traps are located below the workspace in the center of the laboratory. In
TRAPS LITE wasn't my idea. I wanted to do HEAVY METAL TRAPS, a collection of all those SIX-skull traps they wouldn't let me print in my previous four books. I was ready to kick butt and take names. I figure you're ready for it. But, noooooooooo. the publishers are stuck on this TRAPS LITE notion, and there'
need heuristic reasoning when we construct a strict proof as we need scaffolding when we erect a building. . . . Heuristic reasoning is often based on induction, or on analogy. [pp. 112, 1131 Provisional, merely plausible HEURISTIC REASONING is important in discovering the solution, but you should not take it
heuristic relies on familiarity. Based on these results, we created a new theoretical framework that explains decisions attributed to both heuristics based on the underlying memory associated with the choice options. Keywords: recognition heuristic, fluency heuristic, familiarity, recollection, ERPs The study of how people make judgments has .
Description Logic Knowledge Base Exchange Elena Botoeva supervisor: Diego Calvanese PhD Final Examination April 10, 2014 Bolzano Elena Botoeva(FUB)Description Logic Knowledge Base Exchange1/33. Outline 1 Introduction 2 Summary of Work 3 Results 4 Technical Development Universal Solutions Universal UCQ-solutions UCQ-representations Elena Botoeva(FUB)Description Logic Knowledge Base Exchange2/33 .
