Introduction

Learning can be influenced by apparently uncorrelated experience with rewards or punishments. That is, ‘pre-exposure’ to appetitive and aversive stimuli can strongly influence later learning in traditional operant or classical conditioning paradigms. Interestingly, these unpaired experiences can either hinder or enhance later learning. Several phenomena have been identified that reduce later learning, including learned irrelevance and learned helplessness (Seligman 1972; Bennett et al. 1995). An example of this effect is when humans are presented with an unpaired reward and stimulus, their ability to form associative memories later with those same stimuli are strongly reduced (Myers et al. 2000). This effect is not restricted to vertebrate animals. In the honeybee, unpaired sugar or shock exposure retards later associative conditioning (Abramson and Bitterman 1986; Sandoz et al. 2002). These suppressing effects are in contrast to those experiences that can enhance later conditioning. For example, fear conditioning in one context can be behaviorally sensitized when rats experience electric shock in a different context (Rau et al. 2005). Although tail shock in Aplysia can induce memory savings evident in later sensitization experiments (Philips et al. 2006), we know of no evidence that reinforcement pre-exposure can enhance later performance in associative conditioning in an invertebrate animal.

Place conditioning in the heat-box paradigm is ideal for testing whether reinforcement pre-exposure leads to increased or decreased performance in later associative conditioning by Drosophila. In the heat-box, individual flies can walk back and forth in a small narrow chamber lined top and bottom with heating elements (Wustmann and Heisenberg 1997; Zars et al. 2000; Putz and Heisenberg 2002). One half of the chamber can be associated with high temperatures, and during training when a fly moves to one half of the chamber the temperature rises (Zars and Zars 2006). The test of place memory measures persistent place preference in the absence of a rising temperature contingency. In addition to this conditioning procedure, the temperature inside the chamber can be controlled independently of fly behavior. This is usually used to test the ability of a fly to sense and avoid a high-temperature source in the so-called thermosensitivity assay, which can be used to dissociate learning phenotypes of mutant flies from naïve temperature avoidance defects (Zars 2001). We took advantage of the ability to control the temperature inside the chamber to expose flies to unavoidable high temperatures and test for potential effects on later place memory performance.

We investigated here whether unpaired high-temperature exposure can have an effect on later conditioning in Drosophila. To do this, we examined the effect of high-temperature exposures on later conditioning using 41 and 30°C reinforcement. Furthermore, we tested whether the number of high-temperature exposures differentially alters memory performance tested up to 20 min later. Finally, we examined whether changes in the ability to sense and avoid a high-temperature source (thermosensitivity) or the antennae were important for potential pre-exposure effects. From these experiments we found that behaviorally uncorrelated high-temperature exposure can enhance later associative conditioning using reinforcing temperatures that typically do not support memory, and do so independently of changes in temperature perception and the antennal thermosensor. This highlights the perhaps unexpected influence of temperature exposure history on learning that uses high temperatures as negative reinforcement. Furthermore, we provide the first example of enhancement of associative learning after reinforcer pre-exposure in Drosophila, and together with previous results suggest a conserved repertoire of pre-exposure effects in both vertebrate and invertebrate animals.

Materials and methods

Drosophila rearing conditions

We raised wild-type CS flies in our standard cornmeal-based media (Guo et al. 1996) in a 12:12 h light:dark cycle at 25°C and 60% relative humidity. Behavioral experiments were done on 2 to 6-day-old flies. Antennae were removed from flies as in (Zars 2001). Flies that were tested behaviorally were not anesthetized.

Behavioral tests

Heat-box conditioning

Flies were individually trained in a small chamber (Wustmann et al. 1996; Zars et al. 2000). These flies were exposed to a high temperature (either 30 or 41°C) in 1-min blocks. The number of blocks ranged from 0 to 3. In the first experiment (Fig. 1), a 1-min block was followed by a 1 min delay and then conditioning (below). In the latter experiments, high temperature was presented to the flies in a 5 min session, with the three exposure group being presented with 41°C in the first, third, and fifth minutes. In the two (and one) exposure groups, 41°C was presented in the third and fifth (or only the fifth) minute. During the non-exposed periods, the chamber was 24°C. Following a pause of 1–20 min, in which flies remained in their chambers, a conditioning phase began. Conditioning used either 30 or 41°C as negative reinforcement such that when a fly moved to the front (or back) half of the chamber, the whole chamber heated, when that fly moved to the other half of the chamber, it cooled to 24°C. Training duration was 4 min and was followed by a 3 min memory post-test. The 3-min memory performance is presented as a Performance Index (PI). The PI is calculated within a session as the time spent in the ‘safe’ half of the chamber minus the time spent in the punishment-associated half, all divided by the total time in that session. Thus, on this scale, random distribution in the chamber gives a PI of 0, perfect avoidance of the side associated with high temperature gives a 1.0. An equal number of experiments paired high temperature with the back or front half of the chamber, which makes the effect of any potential side preference in the chambers negligible.

Fig. 1
figure 1

Unpaired high-temperature exposure enhances warm temperature reinforced memory. Flies were exposed for 1 min to 24, 30, or 41°C and then conditioned with a 30 or 41°C reinforcing temperature (schematized in upper right panels). a There was no measurable effect of unpaired high-temperature exposure on 41°C reinforced conditioned memory (H (1, N = 144) = 0.554, P = 0.46). In contrast b, flies that had been exposed to 41°C had significantly enhanced memory performance compared to flies exposed to 24 or 30°C (H (2, N = 234) = 15.6, P = 0.0004; multiple comparison tests reveal significant differences between the 41°C group and both 30 and 24°C groups, **P < 0.01). In this and following figs. the bars represent the mean values, error bars are SEMs

Full size image

Thermosensitivity tests

To test for the effects of high-temperature exposure on temperature avoidance behavior, flies were presented with high temperatures in exactly the same fashion as in conditioning tests, but were then tested for their ability to sense and avoid a 30 or 41°C high temperature source. This test used the same chambers as in conditioning, but the temperature in a chamber half was altered independently of a fly’s behavior and alternated chamber halves between test temperatures. The time spent in the cooler and warmer chamber half within a 1-min session was used as above to determine an avoidance PI. Again, equal numbers of experiments were done in which 30 and 41°C were either in the front or back half of the chamber.

Statistics

As tests for normality give mixed results for PIs (Putz and Heisenberg 2002), the non-parametric Kruskal Wallis ANOVA and multiple comparisons of mean ranks for groups were done using the Statistica program (StatSoft, Tulsa, Oklahoma). A sign test against zero was done on the memory performance of flies with and without antennae after reinforcement pre-exposure.

Results

Enhancement of memory performance

In our first attempt to determine whether high temperature exposure could alter later memory performance, we exposed flies to 41°C for 1 min and 60 s later trained them with a short protocol (4 min of training) and tested memory as persistent place preference for 3 min using a 41°C reinforcing temperature. We chose these conditions for the first test because we might reveal either an enhancement or a decrement in performance after pre-exposure since memory performance index (PI) scores with this protocol are typically about 0.5. It should be kept in mind that if flies have no spatial preference in the heat-box they have a PI of 0. If they absolutely avoid the chamber half associated with rising temperatures they have a PI of 1.0. Furthermore, in these and following experiments, we compare the behavior of flies that are pre-exposed to the high temperature reinforcer to those flies kept in the chamber for the same amount of time but with the temperature maintained at the normally preferred 24°C (Sayeed and Benzer 1996; Zars 2001; Zars and Zars 2006). The latter group might detect a context exposure effect. The difference between this context group and the high-temperature group would identify potential reinforcement pre-exposure effects. Our results show that memory performance tested under these conditions is not affected by 41°C pre-exposure (Fig. 1a). Thus, either the 41°C reinforced memory is not altered because flies are performing at levels that cannot be modified or a potential effect depends on pre-exposure to temperatures that are different from the reinforcing temperature. We next exposed flies to either 41, 30, or 24°C for 1 min and tested for place memory after the short conditioning session using 30°C as a negative reinforcer. We chose the 30°C reinforcing temperature because this does not typically support place memories (Zars 2001) and might provide conditions where enhancement could be detected. In these experiments, only flies from the 41°C exposure group had enhanced memory performance (Fig. 1b).

We next asked whether more exposures to high temperatures would further enhance later conditioning. Using the same short protocol with 30°C reinforcing temperature, flies were exposed 0, 1, 2, or 3 times to 41°C, with the interval between the last 41°C exposure and conditioning held constant in all groups (60 s). The time flies spent in the chamber was identical in all groups (330 s). With the 1 min retention interval, flies exposed to 41°C for 1, 2, or 3 times had significantly enhanced conditioned memory performance compared to the flies simply exposed to the chamber but held at 24°C (0-exposures). There did not appear to be an effect of increased number of exposures on memory performance (Fig. 2a).

Fig. 2
figure 2

Unpaired high-temperature enhancement of memory formation lasts at least 20 min. When tested 1 min after exposure, either 1, 2, or 3 exposures enhanced memory formation (a) (protocol schematized in upper right panel; H (3, N = 333) = 13.0, P = 0.005). There was no significant effect of increased number of exposures on memory formation. For both the 10 and 20 min retention tests, 1 and 3 exposures significantly enhanced memory formation (b and c) (10 min retention H (2, N = 216) = 34.6, P = 0.0000; 20 min retention H (2, N = 243) = 30.1, P = 0.0000). Although three exposures increased later memory performance compared to one exposure, this was not significant in either the 10 or 20 min retention tests (multiple comparison tests reveal significant differences only between non-exposed groups and the other groups, *P < 0.05, ***P < 0.001)

Full size image

To provide insights into the duration of the sensitizing effect of 41°C exposure, we exposed three groups of flies either 0, 1, or 3 times to high temperature and tested conditioned memory either 10 or 20 min later (Fig. 2b, c). Under these conditions we found that three exposures had a somewhat larger enhancing effect on memory than one exposure, although this was not statistically significant. Both one and three exposures, however, had a large effect on memory performance compared to no exposures at both 10 and 20 min retention intervals (the time between the high temperature exposure and conditioning). Thus, the enhancing effect of unpaired high temperature exposure lasts at least 20 min.

High temperature exposure does not change unconditioned avoidance behavior and does not require the antennal thermosensor

To determine whether high temperature pre-exposure could alter temperature avoidance behavior, we exposed flies to 41°C under conditions identical to those used above, but tested avoidance of a 30 and 41°C source. In this so-called thermosensitivity assay, flies are presented with a chamber half held at an elevated temperature (30 or 41°C) while the other chamber half is 24°C. Flies normally avoid the chamber half that is higher (or lower) than 24°C, and measures the ability of a fly to both sense and avoid a high temperature source. In all conditions, pre-exposure to 41°C did not have a significant effect on avoidance behavior of either 30 or 41°C (Fig. 3). Thus, the pre-exposure effect on conditioned behavior is independent of changes in high temperature avoidance behavior.

Fig. 3
figure 3

Pre-exposure to high temperature does not alter high-temperature avoidance behavior. With a 1 min retention interval, flies exposed to either 41 or 24 °C showed similar avoidance of 30 and 41°C probe temperatures (procedure schematized in upper right panel; probe temp 30°C, H (1, N = 207) = 0.067, P = 0.79; probe temp 41°C, H (1, N = 207) = 1.21, P = 0.27). Similarly, after a 20 min retention interval, neither one or three 41°C exposures had an effect on avoidance of 30 or 41°C probe temperatures (Single exposure: probe temp 30 °C, H (1, N = 176) = 0.002, P = 0.97; probe temp 41°C, H (1, N = 176) = 1.88, P = 0.17. Three exposures: probe temp 30°C, H (1, N = 173) = 2.53, P = 0.11; probe temp 41°C, H (1, N = 173) = 1.07, P = 0.30)

Full size image

Interestingly, temperatures above and below 30°C appear to be sensed by different mechanisms. Removal of the antennae abolishes flies’ ability to sense and avoid 30°C or less (Sayeed and Benzer 1996; Zars 2001). Furthermore, temperatures above 30°C, are typically required for place memory formation (Zars 2001; Zars and Zars 2006). When the antennae are removed no conditioning has been detected using this temperature (Zars 2001). We, therefore, tested whether the antennal thermosensor was necessary for the pre-exposure effects on memory conditioned with 30°C. To do this, flies with or without antennae were presented with one or three 41°C exposures and tested 20 min later for conditioned place memory using the short protocol and 30°C reinforcement. In both cases, we found that removal of the antennae had no effect on the memory enhancing effect of unpaired high temperature pre-exposure (Fig. 4). Therefore, 41°C exposure recruits the high-temperature thermosensor into responding to 30°C and place memory can be reinforced.

Fig. 4
figure 4

A high-temperature thermosensor is sufficient for reinforcement pre-exposure enhancement of memory formation. Although flies without antennae performed slightly lower than flies with antennae after either one or three exposures (experiment schematized in upper right panel), they did not perform significantly worse in either case (a one exposure H (1, N = 180) = 0.45, P = 0.50, b three exposures H (1, N = 175) = 0.29, P = 0.59). In all groups, memory performance was significantly greater than zero (one exposure with antennae Z = 3.27, P = 0.001; one exposure without antennae Z = 2.2, P = 0.02; three exposures with antennae Z = 4.7, P < 0.001; three exposures without antennae Z = 3.9, P < 0.001)

Full size image

Discussion

Our key finding is that reinforcer pre-exposure can enhance associative memory acquisition in Drosophila. We, thus, provide the first example of this type of enhancement in an invertebrate animal, providing evidence that this component of associative learning is evolutionarily conserved. In place learning, this enhancement of learning is only revealed using reinforcing temperatures that normally require extended experience to alter place memory. Furthermore, we found that the enhancement of place memory is independent of mechanisms that support more straightforward temperature avoidance behavior. This is first indicated with the lack of a change in naïve temperature avoidance following high temperature exposure. And, second, removal of the antennae, normally responsible for sensing 30°C, has no effect on the enhancement of place memory. The high-temperature thermosensor or dependent processes are therefore, likely altered to make 30°C a more effective reinforcing temperature for place memory.

The mechanisms of reinforcement, evident in conditioned behavior, can be dissociated from experience-independent behaviors. For example, conditioning of the proboscis extension reflex (PER) in honeybee is intimately related to activity of the so-called vummx1 neuron (Hammer 1993). However, activity of this neuron cannot by itself elicit the PER. We have also shown previously a separation of temperature avoidance behavior and place learning mechanisms in Drosophila. In this case, we found that mutation of the white-ABC transporter decreased asymptotic memory performance, similar to performance of wild-type flies trained with lower temperatures (Diegelmann et al. 2006). Importantly, the decreased memory level was not a consequence of reduced avoidance behavior of the temperatures used for conditioning. We provide evidence in this paper that a pre-exposing enhancement of memory performance is also independent of changes in naïve temperature avoidance mechanisms.

Two thermosensors have been identified in adult Drosophila (Sayeed and Benzer 1996, Zars 2001). An antennal thermosensor is required for sensing temperatures less than about 30°C. And, removal of the antennae reveals the function of a high-temperature thermosensor. The high temperature thermosensor does not appear to be critical for sensing temperature below 30°C, but is important for sensing temperatures above this threshold. Recruitment of the high temperature thermosensor is necessary for place learning in the heat-box using the short protocol, but with extended training 30°C can be used to support limited memory performance (which presumably functions through the antennal thermosensor) (Zars 2001; Zars and Zars 2006). Clues about the anatomical nature of the high-temperature thermosensor come from investigation of two TRP channels, trpA1 and painless. In the larval stage, these two TRP channels have been identified as critical for sensing high temperatures (Tracey et al. 2003; Zars 2003; Rosenzweig et al. 2005). And, the painless TRP is necessary for 47°C response in the adult fly (Xu et al. 2006). Whether these channels are important for temperature avoidance and conditioned behavior in the heat-box is still open. Regardless of the questions on the nature of the high-temperature thermosensor, our results support the notion that the mechanisms of the pre-exposure effect act upon the high-temperature thermosensor or on downstream systems to enhance memory formation.

Enhancement of associative learning by reinforcement pre-exposure could be accomplished by increasing the sensitivity of the peripheral sensory afferent pathway, ‘inflation’ of high-temperature reinforcing value, altering memory formation mechanisms, or activation of another system, which can influence the associative process. In the rat, it appears that sensitization can influence all of these processes except at the memory forming level (Rau et al. 2005). For example, pain processes can be enhanced with electric shock exposure (King et al. 1996). Furthermore, reinforcement value can be increased with experience (Bouton 1984). And, in the enhancement of fear conditioning in different contexts, it seems that pre-exposure to electric shock can enhance new context memory by altering a stress response system (Rau et al. 2005). We show here that associative learning in the fly can be enhanced and does so independently of changes in the sensory afferent pathway. Further experiments, including tests of different contexts, will determine whether reinforcement pre-exposure alters memory at any of the other three potential levels in the fly.