**2. Avoidance as an alternative symptom to model**

#### **2.1 Behavioral avoidance**

Avoidance, though not as well studied as hyperarousal or reexperiencing, is a common symptom to all anxiety disorders, including PTSD. Recent research indicates that the presentation of increased avoidant behavior has been found to track the general worsening of PTSD symptoms (Karamustafalioglu et al., 2006; O'Donnell et al., 2007) suggesting that acquisition of avoidance may have an etiological role in the development of PTSD. Many animal models of avoidance use tasks that are based on a tendency to exhibit passive avoidance. Tasks such as the elevated plus maze, measure how often rodents explore elevated open-arms (no walls) versus arms with high walled sides (closed arms). Rodents have a fundamental aversion to well-lit open spaces; therefore, prior exposure to stressors and anxiogenic drugs will reduce the limited tendency to explore the open arms (Pellow et al., 1985); conversely, anxiolytics increase exploration into the open arms. However, avoidance symptoms displayed in PTSD get progressively worse over time, and models such as the plus maze are not conducive to exhibit such a progression. In fact, avoidance and avoidant behaviors distinguish between those that develop PTSD and those who recover from trauma (Karamustafalioglu et al., 2006; Foa et al., 2006; O'Donnell et al., 2007). Thus the adoption of behavioral, cognitive, or emotional avoidance represents the detrimental process, which distinguishes those that successfully cope from those that develop pathological anxiety. Further, many avoidance behaviors in humans are active, that is, behaviors are not merely the absence of activity. Active avoidance is often more debilitating as the increased time and resources utilized for avoidance limit the capability of the individual to perform other tasks. Therefore, a task that allows for the observance of a methodical increase in active avoidant behaviors would be more akin to what is described in PTSD.

The process of learning active avoidance behaviors in rodents involves three steps. One, the rodent has to learn how to instrumentally dissociate itself from a noxious stimulus. This either may involve removing itself from the noxious stimulus (as in shuttle escape) or manipulating the environment such that the presentation of the noxious stimulus stops (as in lever-press escape or wheel-turn). Two, the rodent needs to recognize that the presence of certain stimuli in the environment precede the presentation of the noxious stimulus. These "warning signals" need to be distinct from the environment, with auditory stimuli serving as better signals than visual stimuli (Gilbert, 1971). This association may occur either during or following the acquisition of escape behavior. Three, the rodent needs to use the warning signal as a cue to emit an instrumental response prior to the actual onset of the noxious stimulus (i.e. avoidance behavior). The emitting of the response to the predictive warning signal removes the warning signal and the associated threat. In some regard, this process is more complicated for the rodent because it usually has to learn an escape response before it will reliably learn an avoidance response; although for some the transition from escape to avoidance behavior may involve less training sessions than for others. Humans can obviously learn to avoid people, animals, places, and situations without necessarily having to learn to escape from them first, but because this learning process is a bit more methodical in the rodents, we can understand how each stage of the process occurs and how it contributes to pathological avoidance.

Pathological avoidance is when the animal (human or non-human) responds to warning signals nearly 100% of the time. Although intuitively this may seem to be a responsible

Avoidance, though not as well studied as hyperarousal or reexperiencing, is a common symptom to all anxiety disorders, including PTSD. Recent research indicates that the presentation of increased avoidant behavior has been found to track the general worsening of PTSD symptoms (Karamustafalioglu et al., 2006; O'Donnell et al., 2007) suggesting that acquisition of avoidance may have an etiological role in the development of PTSD. Many animal models of avoidance use tasks that are based on a tendency to exhibit passive avoidance. Tasks such as the elevated plus maze, measure how often rodents explore elevated open-arms (no walls) versus arms with high walled sides (closed arms). Rodents have a fundamental aversion to well-lit open spaces; therefore, prior exposure to stressors and anxiogenic drugs will reduce the limited tendency to explore the open arms (Pellow et al., 1985); conversely, anxiolytics increase exploration into the open arms. However, avoidance symptoms displayed in PTSD get progressively worse over time, and models such as the plus maze are not conducive to exhibit such a progression. In fact, avoidance and avoidant behaviors distinguish between those that develop PTSD and those who recover from trauma (Karamustafalioglu et al., 2006; Foa et al., 2006; O'Donnell et al., 2007). Thus the adoption of behavioral, cognitive, or emotional avoidance represents the detrimental process, which distinguishes those that successfully cope from those that develop pathological anxiety. Further, many avoidance behaviors in humans are active, that is, behaviors are not merely the absence of activity. Active avoidance is often more debilitating as the increased time and resources utilized for avoidance limit the capability of the individual to perform other tasks. Therefore, a task that allows for the observance of a methodical increase in active avoidant behaviors would be more akin to what is described in

The process of learning active avoidance behaviors in rodents involves three steps. One, the rodent has to learn how to instrumentally dissociate itself from a noxious stimulus. This either may involve removing itself from the noxious stimulus (as in shuttle escape) or manipulating the environment such that the presentation of the noxious stimulus stops (as in lever-press escape or wheel-turn). Two, the rodent needs to recognize that the presence of certain stimuli in the environment precede the presentation of the noxious stimulus. These "warning signals" need to be distinct from the environment, with auditory stimuli serving as better signals than visual stimuli (Gilbert, 1971). This association may occur either during or following the acquisition of escape behavior. Three, the rodent needs to use the warning signal as a cue to emit an instrumental response prior to the actual onset of the noxious stimulus (i.e. avoidance behavior). The emitting of the response to the predictive warning signal removes the warning signal and the associated threat. In some regard, this process is more complicated for the rodent because it usually has to learn an escape response before it will reliably learn an avoidance response; although for some the transition from escape to avoidance behavior may involve less training sessions than for others. Humans can obviously learn to avoid people, animals, places, and situations without necessarily having to learn to escape from them first, but because this learning process is a bit more methodical in the rodents, we can understand how each stage of the process occurs and how it

Pathological avoidance is when the animal (human or non-human) responds to warning signals nearly 100% of the time. Although intuitively this may seem to be a responsible

**2. Avoidance as an alternative symptom to model** 

**2.1 Behavioral avoidance** 

PTSD.

contributes to pathological avoidance.

strategy for the animal, it does not allow for the individual to be sensitive to contingency changes (i.e. when the warning signal no longer reliably predicts the noxious stimulus). At that point, the avoidance responses are being emitted to remove a possible (not probable) threat. Therefore, the animal may be expending energy, by moving to the lever and subsequently depressing it, trial after trial to avoid a threat that the warning signal no longer reliably predicts. For individuals with severe anxiety disorders, this strategy of avoiding possible threats can become very disruptive if 1) the individuals expend more energy to avoid situations than what would be required to actually deal with them and 2) the perceived warning signals become generalized, which narrows their ability to interact with the world. Therefore, identifying an animal model that will acquire an exceptionally high asymptotic level of avoidant behavior, and subsequently exhibits the predictable slow extinction of the response, can provide us with a valuable system for identifying the vulnerability factors that predict such avoidant behaviors as well as the neural mechanisms that bias their behavioral strategies to such an extreme.

There are various forms of active avoidance that can be modeled in rats, but the desire to track the development of increased avoidant behavior over time led us to adopt distinct lever-press avoidance as our active avoidance procedure. Lever-press avoidance has been utilized for decades to study learning, but it also has a history as a prominent model of anxiety (Pearl, 1963; D'Amato & Fazzaro, 1966; Hurwitz & Dillow, 1968; Gilbert, 1971; Dillow et al., 1972; Berger & Brush, 1975). Derived initially from the 2-factor theory of threat/fear motivation and learned avoidance (Mowrer, 1939a; Mowrer, 1939b; Mowrer & Lamoreaux, 1942; Mowrer & Lamoreaux, 1946), the general premise of this approach is that a learned fear of signals is sufficient to support avoidant behavior without requiring a continued re-exposure to the actual noxious stimulus or event. Others have provided alternative interpretations of the development of active avoidance learning. Herrnstein, Hineline, and Sidman all focused on the reduction in shock density over time and a second internal factor (e.g fear or anxiety) need not be required in order to explain the acquisition of avoidance behavior (Sidman, 1962; Herrnstein & Hineline, 1966; Hineline & Herrnstein, 1970). This is an important consideration, for without the theoretical need for an internal state, there is no reason to assume a general state of arousal should be evident in the absence of shock exposure. In short, once asymptotic performance is attained, because of the adaptation of the instrumental response to minimize shock frequency, general arousal should be reduced compared to early acquisition (when shocks are more frequent). Still, others have suggested that there may be another component to this acquired behavior – the attainment of perceived safety (Dinsmoor, 1977; Dinsmoor, 2001). This is an interesting proposition because it also does not require any rumination upon the animal's part to "know" the shock is coming. In this approach, the animal learns to exhibit the behavior because it leads to the attainment of perceived safety, which could be in the form of an explicit stimulus only present during periods of non-threat or simply as the absence of the warning signal. At the foundation of this theoretical discussion is a fundamental difference in the view of how animals perceive learning: a molecular (trial by trial, stimulus by stimulus) or molar (general state) analysis (Hineline, 2001; Bersh, 2001). One could argue that lingering changes in general arousal outside of the avoidance learning context may reflect overall changes in the animals that would be proposed by molar analysis theory.

## **2.2 Avoidance susceptibility as a model of anxiety vulnerability**

As mentioned above, it is well documented that approximately 10% of those people who experience a significant trauma develop PTSD; therefore, there has been recent interest in

Acquisition of Active Avoidance Behavior as a Precursor

1968; Forsyth, 1969).

coping strategies strengthen.

to Changes in General Arousal in an Animal Model of PTSD 79

For some time, the acquisition and performance of avoidance behavior was used as a tool to increase arousal in studies of physiological responsiveness in monkeys (e.g. stress-induced hypertension) because it was obvious to the investigators that control over the stimuli did not necessarily lead to a reduction in arousal (Forsyth, 1968; Forsyth, 1969; Natelson et al., 1976; Natelson et al., 1977). Since that time, avoidance learning fell out of favor as such a tool and was generally replaced by inescapable stressor paradigms. Therefore, with our rodent model of acquisition and extinction of active avoidance behavior, we questioned whether the process of acquiring avoidant behavior would influence general arousal outside of the avoidance-training context, which was not the case for some of the monkey studies (Forsyth,

Having established strain and sex differences in the acquisition and extinction of active avoidance, as well as differences in innate reactivity between strains, the question became whether the process of acquiring avoidant behavior would influence general arousal outside of the avoidance-training context. There are three possible periods of time startle reactivity may show changes as a function of acquiring lever-press avoidance and each would have associated with it a different theory of how the learning procedure was affecting general sensory reactivity. First, based on the above inescapable shock model, one could hypothesize that startle reactivity should be increased within days of the first few training trials, following the sessions the rats experience the most shock. Second, if the development of avoidant behavior follows the trajectory of developing anxiety, then one could hypothesize that startle reactivity should increase over acquisition. Yet, there is also a third option. That is, startle reactivity could increase if the association between the signals and the consequence becomes less certain. In this third possibility, startle reactivity could be increased if there is a change in the relationship between the signals that represent threat and the consequences following acquisition (such as conducting extinction trials). Another consideration is that only certain animals may be affected in a way that increases their general arousal. Strain differences in both acquiring the avoidant behavior and resistance to extinguish it may be a sign of anxiety vulnerability that could also be reflected in a change in

**3. Behavioral avoidance as a precursor for increased arousal** 

general arousal (reflected as a persistent change in startle reactivity).

**3.1 Acquisition of active avoidance and changes in startle reactivity** 

There are several examples of shock-induced changes in various behavioral indexes of anxiety-like reactions outside of the shock-exposure context (Servatius et al., 1994; Servatius et al., 1995; Beck et al., 2002; Cordero et al., 2003; Beck & Servatius, 2005; Manion et al., 2007; Daviu et al., 2010; Manion et al., 2010), which may lead one to assume that stressors which cause pain have a particularly significant role in causing context-independent changes in general arousal. However, in the case of acquiring behavior that is conducive to active avoidance of shock, the acute role of shock exposure in early acquisition can be contrasted with the expression of stimulus control during asymptotic performance levels. This is an important distinction in any avoidance-based model of PTSD, since the clinical condition does not necessarily involve an acute increase in arousal (as defined by startle reactivity), but the development of avoidance does parallel the general worsening of symptoms (O'Donnell et al., 2007). The implication of this correlation is that other symptoms, such as hyperarousal, may come as a result of increasing stimulus control, as active avoidance

identifying vulnerability factors that cause some proportion of the public to be susceptible to developing PTSD symptoms. From a learning-diathesis approach, vulnerability for developing anxiety disorders comes from differences in acquiring associations. People selfascribed as being behaviorally inhibited, as well as the rat model of behavioral inhibition (the WKY rat), exhibit quicker acquisition of classically conditioned responses (Ricart et al., 2011b; Myers et al., 2011; Beck et al., 2011). In addition, females also exhibit enhanced susceptibility to acquire associations. This is reflected in faster acquisition of predictive relationships (classical conditioning) (Spence & Spence, 1966; Wood & Shors, 1998; Shors et al., 1998; Holloway et al., 2011) and behavioral reactions to stimuli (instrumental learning) (Van Oyen et al., 1981; Heinsbroek et al., 1983; Heinsbroek et al., 1987; Saavedra et al., 1990; Dreher et al., 2007; Dalla et al., 2008; Lynch, 2008), the 2 primary components of avoidance learning (Mowrer & Lamoreaux, 1946). Based on these characteristics, it is not surprising that both female sex and behaviorally inhibited temperament are associated with a greater susceptibility to acquire active avoidance behaviors (Beck et al., 2010; Beck et al., 2011). As shown in Figure 3, male SD rats are slowest to acquire a lever-press avoidance response, compared to their same-strain female counterparts and WKY rats of both sexes. Interestingly, the relationships between the 4 groups change during extinction with male WKY rats extinguishing slower than both female groups and male SD rats. Females of both strains in this study both acquired and extinguished at the same rate.

Fig. 3. Avoidance susceptibility can be observed by comparing rates active avoidance responses are acquired. In this example, WKY rats acquire a lever-press avoidance behavior quicker than male SD rats, main effect Strain F (1, 36) = 9.0, p < .005. Female SD rats also acquired the behavior quicker than male SD rats, Sex x Session F (9, 324) = 2.6, p < .01. Following session 10, the shock was removed in order to assess extinction of the avoidance response. In general, WKY rats extinguish the response slower than SD rats, but female SD rats were slower than male SD rats extinguishing the behavior, whereas male WKY rats were slower to extinguish the response compared to female WKY rats, Strain x Sex F (1, 36) = 6.0, p < .02 and Sex x Session F (5, 179) = 2.2, p < .05.

identifying vulnerability factors that cause some proportion of the public to be susceptible to developing PTSD symptoms. From a learning-diathesis approach, vulnerability for developing anxiety disorders comes from differences in acquiring associations. People selfascribed as being behaviorally inhibited, as well as the rat model of behavioral inhibition (the WKY rat), exhibit quicker acquisition of classically conditioned responses (Ricart et al., 2011b; Myers et al., 2011; Beck et al., 2011). In addition, females also exhibit enhanced susceptibility to acquire associations. This is reflected in faster acquisition of predictive relationships (classical conditioning) (Spence & Spence, 1966; Wood & Shors, 1998; Shors et al., 1998; Holloway et al., 2011) and behavioral reactions to stimuli (instrumental learning) (Van Oyen et al., 1981; Heinsbroek et al., 1983; Heinsbroek et al., 1987; Saavedra et al., 1990; Dreher et al., 2007; Dalla et al., 2008; Lynch, 2008), the 2 primary components of avoidance learning (Mowrer & Lamoreaux, 1946). Based on these characteristics, it is not surprising that both female sex and behaviorally inhibited temperament are associated with a greater susceptibility to acquire active avoidance behaviors (Beck et al., 2010; Beck et al., 2011). As shown in Figure 3, male SD rats are slowest to acquire a lever-press avoidance response, compared to their same-strain female counterparts and WKY rats of both sexes. Interestingly, the relationships between the 4 groups change during extinction with male WKY rats extinguishing slower than both female groups and male SD rats. Females of both

strains in this study both acquired and extinguished at the same rate.

Fig. 3. Avoidance susceptibility can be observed by comparing rates active avoidance responses are acquired. In this example, WKY rats acquire a lever-press avoidance behavior quicker than male SD rats, main effect Strain F (1, 36) = 9.0, p < .005. Female SD rats also acquired the behavior quicker than male SD rats, Sex x Session F (9, 324) = 2.6, p < .01. Following session 10, the shock was removed in order to assess extinction of the avoidance response. In general, WKY rats extinguish the response slower than SD rats, but female SD rats were slower than male SD rats extinguishing the behavior, whereas male WKY rats were slower to extinguish the response compared to female WKY rats, Strain x Sex F (1, 36)

= 6.0, p < .02 and Sex x Session F (5, 179) = 2.2, p < .05.
