Because various types of experience can affect organizational learning processes and outcomes differently, researchers have characterized experience at a fine-grained level along various dimensions (Argote, McEvily, & Reagans, 2003). The most fundamen- tal dimension of experience is whether it is acquired directly by the focal organiza- tional unit or indirectly from other units (Argote, 2012). Learning from the latter type of experience is referred to as vicarious learning (Bandura, 1977) or knowledge trans- fer (Argote, Ingram, Levine, & Moreland, 2000), which is discussed in Chap. 6.
A unit of task experience can also be characterized in terms of its novelty, success, ambiguity, timing, and geographic location. The cumulative amount of experience can be characterized in terms of its heterogeneity and pace. Argote and Todorova (2007) reviewed the effects of different types of experience on learning process and outcomes. Major findings and recent developments are highlighted here.
1. Direct Versus Indirect Experience
Early learning curve studies investigated how organizations learn from their own direct experience (see Yelle, 1979, for a review). Although the rate of learning has been found to vary across organizations, considerable evidence that organizations learn from their own direct experience has accumulated (Dutton & Thomas, 1984). More recently, researchers have investigated how organizational units learn from the experience of other units (Darr et al., 1995; Szulanski, 1996). This latter form of learning is also referred to as knowledge transfer.
An important research issue is the relationship between direct and indirect experience. Several researchers have found that direct experience and indirect expe- rience are negatively related (Haas & Hansen, 2005; Schwab, 2007; Wong, 2004). That is, one form of experience seems to substitute for the other. By contrast, other researchers have found that direct and indirect experience relate positively to each other in complementary fashion (Bresman, 2010). Understanding the conditions under which direct and indirect experience complement or substitute for each other is an important question that would benefit from further research.
2. Novelty of Experience
Experience can be acquired on a novel task or on a task that has been performed repeatedly in the past. March (1991) distinguished between “exploitation,” which involves learning from repeating the same tasks (low novelty), and “exploration,” which involves learning from new tasks (high novelty). Researchers have investi- gated the relationship between exploitation and exploration. Although originally conceived as a trade-off, exploitation and exploration have been found to be inde- pendent dimensions in several studies (Katila & Ahuja, 2002). There is considerable evidence that organizations need to both explore and exploit in order to be effective (He & Wong, 2004; Katila & Ahuja, 2002; Knott, 2001). Research on “organiza- tional ambidexterity” investigates how organizations can both explore and exploit (see Raisch, Birkinshaw, Probst, & Tushman, 2009, for a review).
3. Success Versus Failure Experience
A unit of task experience can be a success or a failure. Organizations learn from both successes and failures. Denrell and March (2001) argued that learning pro- cesses are biased because of the tendency of individuals to sample and replicate successful experiences. Organizations can learn from failed units of experience. For example, Haunschild and Sullivan (2002) found that airlines learned from acci- dents, a failure in their context. Similarly, Baum and Dahlin (2007) found that prior accident experience reduced the costs of future accidents reported by US railroads and Madsen (2009) found that organizations in the coal mining industry learned from their own accidents and the accidents of other firms.
Sitkin (1992) proposed that learning from failure is more effective than learning from success because failure motivates deeper search and richer understandings than success. Consistent with this argument, Madsen (2009) found in his study of accidents in coal mines that the effect of minor accident experience decayed at a faster rate than the effect of disaster experience, major accidents in which lives were lost. Similarly, Madsen and Desai (2010) found that knowledge acquired from failure experience decayed more slowly than knowledge acquired from success experience in their study of orbital launches.
Other studies, however, have found that organizations learn more from success than from failure or learn from both success and failure. For example, Gino, Argote, Miron-Spektor, and Todorova (2010) found that laboratory teams learned more from other teams that developed a successful product than from other teams that developed an unsuccessful one. In a study of chains of nursing homes, Chuang and Baum (2003) found that organizations learned both from their own failures and from the failures of other organizations but that they learned less from their own failures when the organization was invested in the failed activity. Differences in motivation may reconcile these disparate findings on learning from failure. When the failure is very serious such as an airline (Haunschild & Sullivan, 2002), mining (Madsen, 2009), or orbital launch (Madsen & Desai, 2010) accident, organizations are very motivated to learn from the failures. On the other hand, if the stakes are not very high or if organizations are invested in the failed activity (Chuang & Baum, 2003), learning from failure occurs less frequently.
Learning from contrasting successful and unsuccessful experiences can be especially effective. Kim, Kim, and Miner (2009) found that learning occurred from both success and failure experience, at least after a threshold level of experience was obtained. Further, success and failure experience operated as complements, enhanc- ing each other’s value.
4. Ambiguity of Experience
Experience can be ambiguous (March, 2010) or easily interpretable. Causally ambig- uous experience occurs when the relationship between causes and effects during task performance is unclear. Causal ambiguity makes it hard to interpret experience (Bohn, 1994; Carley & Lin, 1997) and can lead to “superstitious” learning (Levitt & March, 1988) in which participants draw the wrong inferences from experience.
Delays between actions and their effects contribute to causal ambiguity. Diehl and Sterman (1995) found that participants did not learn much from experience when delays between causes and effects occurred. Similarly, Repenning and Sterman (2002) found that in contexts where there were delays between making a process improvement and observing results, participants made attribution errors about the causes of results.
5. Spatial Location of Experience
An organization’s experience can be geographically concentrated or geographi- cally dispersed (Cummings, 2004; Gibson & Gibbs, 2006). Learning from geographically distributed experience poses challenges to organizational learning but also provides opportunities for accessing new knowledge (Argote, Denomme, & Fuchs, 2011). Organizational units that are geographically dispersed have access to more knowledge than those that are geographically concentrated (Ahuja & Katila, 2004). Geographically distributed units, however, face challenges exchanging infor- mation and are more likely to encounter motivational and relational problems than collocated units (Cramton, 2001). Relative to geographically distributed unit, geographically collocated units are more likely to develop “common ground” (Fussell & Krauss, 1992) or shared understandings that facilitate information exchange and the interpretation of experience.
6. Timing of Experience
Experience can be characterized along several temporal dimensions, including its timing and recency. Experience can be acquired before doing, through activities such as training or experimentation (Carrillo & Gaimon, 2000; Pisano, 1994). Experience can be acquired during task performance through learning by doing. Experience can also be acquired after task performance through “after action” reviews (Ellis & Davidi, 2005).
The most effective timing of experience depends on the extent to which cause– effect relationships are understood and the knowledge base in an area is developed. Pisano (1994) found that if the knowledge base was well understood, experimenta- tion and learning before doing contributed to more rapid product development. By contrast, if the knowledge base was not well understood, laboratory experimenta- tion did not advance product development. Similarly, Eisenhardt and Tabrizi (1995) found that learning by doing was more effective for launching new computer prod- ucts than planning or learning before doing was.
Recency is another dimension along which experience can vary. A unit of task experience could have been acquired recently or it could have been acquired in the distant past. There is considerable evidence that recent experience is more valuable than experience acquired in the distant past. That is, experience appears to decay or depreciate (Argote, Beckman, & Epple, 1990; Benkard, 2000; Darr et al., 1995). Further, the rate of depreciation varies across organizations with some organizations showing rapid deprecation and others showing little or no deprecation. The causes of depreciation are discussed in Chap. 3.
7. Rareness of Experience
Experience can vary in its frequency (Herriott, Levinthal, & March, 1985; Levinthal & March, 1981). Experience that occurs rarely or infrequently is hard to interpret and thus poses challenges to learning (Lampel, Shamsie, & Shapira, 2009; March, Sproull, & Tamuz, 1991). Rare experience can lead organizations to draw the wrong inferences from experience and engage in superstitious learning (Zollo, 2009). Organizations, however, can realize significant benefits from learning from rare events (Starbuck, 2009), especially when they invest in developing lessons to improve how they respond to rare events in the future (Rerup, 2009). Further, rare events can interrupt routine activity and extend an organization’s understanding of its capabilities and identity (Christianson, Farkas, Sutcliffe, & Weick, 2009).
8. Simulation of Experience
A dimension related to the timing of experience is the extent to which the experi- ence is simulated. Simulated experience typically occurs after or before—but not during—task performance. Simulated experience might occur before performing the task through “preparedness drills” in which members practice their roles. Computational models that simulate how members and tools interact to perform tasks under various contextual conditions can also be used to facilitate learning before doing the task. Another form of simulated experience is produced through counterfactual thinking (Morris & Moore, 2000; Roese & Olson, 1995). Counterfactual thinking, which typically occurs after doing, involves reconstruction of past events and consideration of alternatives that might have occurred.
The usefulness of simulated experience depends on the extent to which it captures relevant features of the task performance context. Simulated experience can be a valuable complement to real experience, especially when that experience is sparse and/or the stakes are high. For example, disaster drills are conducted at hospitals to enable staff to handle real disasters effectively. Simulations can be especially valu- able in revealing relationships and interactions among the elements of a task perfor- mance system.
The dimensions of experience discussed thus far can refer to a particular unit of experience or can refer to the overall distribution of experience when aggregated. For example, a particular unit of experience can be acquired in a collocated or geo- graphically distributed fashion and the overall spatial distribution of cumulative experience can be obtained by aggregating the experience of particular units. Other dimensions, such as heterogeneity, make sense only as characterizations of cumula- tive experience. These dimensions are now discussed.
9. Heterogeneity of Experience
An organization’s overall distribution of task experience can be characterized in terms of its heterogeneity. Organizations performing similar tasks would be low in heterogeneity while organizations performing varied tasks would be high in heterogeneity. Several studies suggest that some heterogeneity or diversity in task experience facilitates organizational learning. Littlepage, Robison, and Reddington (1997) found that experience on related tasks enhanced group learning by increas- ing individual members’ proficiency, while experience on comparable but not related tasks enhanced learning by increasing members’ knowledge of who was good at which tasks. Similarly, Schilling, Vidal, Ployhart, and Marangoni (2003) found that related task experience improved learning to a greater extent than either identical or unrelated task experience. Boh, Slaughter, and Espinosa (2007) found that diverse experience in related systems improved the performance of software teams. Heterogeneity appears to be most valuable when it is introduced slowly so that members have time to master the different tasks (Pisano, Bohmer, & Edmondson, 2001).
Focusing on heterogeneity of outcomes, Haunschild and Sullivan (2002) found that heterogeneous accident experience was more conducive to organizational learn- ing than homogeneous experience for specialized airlines. Kim et al. (2009) found that success and failure experience enhanced each other’s effect on learning. Zollo (2009) found in a study of corporate acquisitions that heterogeneous acquisition experience weakened the negative effect of past success on the performance of a focal acquisition. Thus, heterogeneous experience reduced the likelihood of super- stitious learning. Some degree of heterogeneity in task outcomes enhances learning by providing organizational members with a deeper understanding of what contrib- uted to successful task performance.
10. Pace of Experience
Another temporal dimension along which experience can vary is its pace. Organizations can acquire experience at a steady rate or they can acquire experience at an uneven rate, with interruptions in production. Interruptions can lead to knowl- edge decay or depreciation (Argote et al., 1990; Benkard, 2000). Interruptions also provide opportunities for knowledge transfer (Zellmer-Bruhn, 2003).
Source: Argote Linda (2013), Organizational Learning: Creating, Retaining and Transferring Knowledge, Springer; 2nd ed. 2013 edition.