1.1 Transfer of the dopamine response from reward to reward-predicting cues

Extensive human and animal research over the past few decades has revealed a central role for dopamine in positive reinforcement. In non-human primates and rodents, experiments have shown that dopamine neurons in the midbrain are activated by unexpected primary rewards, such as sips of juice or pieces of apple delivered at random times (Ljungberg, Apicella, & Schultz, Reference Ljungberg, Apicella and Schultz1992; Schultz, Apicella, & Ljungberg, Reference Schultz, Apicella and Ljungberg1993; Schultz, Dayan, & Montague, Reference Schultz, Dayan and Montague1997). However, when these primary rewards are repeatedly preceded by a sensory stimulus (such as a sound tone or light), the response of the dopamine neurons to the reward decreases, and the response to the associated sensory stimuli increases (Day, Roitman, Wightman, & Carelli, Reference Day, Roitman, Wightman and Carelli2007; Ljungberg et al., Reference Ljungberg, Apicella and Schultz1992; Pan, Schmidt, Wickens, & Hyland, Reference Pan, Schmidt, Wickens and Hyland2005; Schultz et al., Reference Schultz, Apicella and Ljungberg1993, Reference Schultz, Dayan and Montague1997). After repeated pairing with reward, the previously neutral sensory stimuli become “reward-predicting cues” and cause dopamine release, while over the same period, dopamine release in response to the actual reward decreases. We refer to this as “transfer” of the dopamine response from reward to reward-predicting cues (see Fig. 1a).

Figure 1. Transfer of dopamine response from actual reward to cues predicts behavioral characteristics. Traces show idealized dopamine signal in normal (a) and hypothesized dopamine transfer deficit (b). In both cases, unexpected primary reward causes a dopamine response. Normally, after repeated pairing of cue and reward, the dopamine response transfers to the cue. When there is a dopamine transfer deficit the cue response fails to develop as strongly as normal, and the response to the actual reward persists. Compared to normal rat strains, the SHR shows a dopamine transfer deficit. This is associated with characteristic of immediate over-delayed reward. In humans, a dopamine transfer deficit may give rise to symptoms of ADHD. These can be viewed as extremes of normal variations in individual personality traits.

1.2 Actions of dopamine at the behavioral level

Without knowing what its effects are, knowledge about the release of dopamine and its transfer from rewards to cues has little explanatory power. Several distinguishable, but not mutually exclusive, hypotheses have been proposed for the actions of dopamine, namely: positive reinforcement, reward, hedonia, incentive motivation, and conditioned reinforcement (Wise, Reference Wise2004).

The most established hypothesis is that dopamine mediates positive reinforcement, which by definition increases the frequency of responses that it follows. Depleting dopamine or blocking its action reduces the reinforcing effects of food, water, or direct stimulation of dopamine cells (Beninger & Freedman, Reference Beninger and Freedman1982; Lynch & Wise, Reference Lynch and Wise1985; Wise, Reference Wise2006). Under such conditions, learning new responses is prevented, previously trained responses decline in frequency, and acquisition of Pavlovian stimulus-reward associations is reduced (Darvas, Wunsch, Gibbs, & Palmiter, Reference Darvas, Wunsch, Gibbs and Palmiter2014). Conversely, stimulating the dopamine-rich areas of the brain (ventral tegmental area and substantia nigra pars compacta of the midbrain and their projections to the nucleus accumbens and dorsal striatum) is sufficient to cause positive reinforcement (Corbett & Wise, Reference Corbett and Wise1980; Kim et al., Reference Kim, Baratta, Yang, Lee, Boyden and Fiorillo2012; Wise & Bozarth, Reference Wise and Bozarth1984). Thus the dopamine neurons of the midbrain are both necessary and sufficient for positive reinforcement of responses.

The positive reinforcement hypothesis of dopamine action suggests that dopamine acts on the memory traces of past responses to increase their frequency in the future. The reward hypothesis, on the other hand, refers to proactive effects of dopamine on future actions. “Reward” in this context refers to stimuli that evoke approach to the stimulus. Dopamine release energizes future responses and approaches to behavior (Gallistel, Stellar, & Bubis, Reference Gallistel, Stellar and Bubis1974). Dopamine is not necessary for producing these responses but increases their vigor or reduces their latency (Wise, Reference Wise2004).

A third effect of dopamine is related to incentive motivation. This effect refers to the way sensory stimuli can acquire incentive properties or value by prior association with reward (Berridge, Reference Berridge2000). Sensory stimuli that have developed incentive properties can powerfully control behavior (Bindra, Reference Bindra1974). Such stimuli elicit orientation and approach toward them. They can also produce positive reinforcement in the absence of actual reward, in which case they are considered conditioned reinforcers. The transfer of dopamine release from the time of actual reward to the time of the reward-predicting cue may be what gives these sensory stimuli incentive value.

Another theory of dopamine function concerns its role in pleasure. The dopamine hedonia theory is less well-supported. Although rewards are a source of pleasure, elevations of brain dopamine are not strongly correlated with subjective pleasure (Berridge, Reference Berridge2000, Reference Berridge2007). On the other hand, decreased dopamine release in the striatum in response to rewards is associated with reduced ability to experience pleasure (anhedonia) in depression (Belujon & Grace, Reference Belujon and Grace2017; Phillips et al., Reference Phillips, Walsh, Zurcher, Lalush, Kinard, Tseng and Dichter2023) and there is evidence that dopamine receptors are important in the action of antidepressants (Willner, Hale, & Argyropoulos, Reference Willner, Hale and Argyropoulos2005).

These theories concerning the action of dopamine are at the level of behavior and subjective experience. However, it is also important to consider the underlying neural mechanisms by which dopamine produces these effects and the requirements for activation of those mechanisms. In broad anatomical terms, the dopamine neurons of the midbrain project to cortical and subcortical areas, most densely to the dorsal striatum, and to a lesser extent, the ventral striatum of the basal ganglia (Swanson, Reference Swanson1982). At the cellular level, dopamine neurons make synaptic contacts on the same neurons and at the same location as the excitatory inputs to those regions (Smith, Bennett, Bolam, Parent, & Sadikot, Reference Smith, Bennett, Bolam, Parent and Sadikot1994) indicating that dopamine acts as a modulator of other synaptic inputs.

1.3 Actions of dopamine at the cellular level

Many pieces of evidence indicate that the positive reinforcement effects of dopamine are mediated by strengthening of synaptic connections between the cerebral cortex and the striatum. Dopamine facilitates long-term potentiation of these corticostriatal synaptic connections under certain conditions (Calabresi, Picconi, Tozzi, & Di Filippo, Reference Calabresi, Picconi, Tozzi and Di Filippo2007; Centonze, Picconi, Gubellini, Bernardi, & Calabresi, Reference Centonze, Picconi, Gubellini, Bernardi and Calabresi2001; Pawlak & Kerr, Reference Pawlak and Kerr2008; Reynolds et al., Reference Reynolds, Hyland and Wickens2001; Shen, Flajolet, Greengard, & Surmeier, Reference Shen, Flajolet, Greengard and Surmeier2008; Shindou et al., Reference Shindou, Shindou, Watanabe and Wickens2019; Wickens, Reference Wickens2008; Wickens, Begg, & Arbuthnott, Reference Wickens, Begg and Arbuthnott1996; Yagishita et al., Reference Yagishita, Hayashi-Takagi, Ellis-Davies, Urakubo, Ishii and Kasai2014). At the cellular level, a three-term contingency has been demonstrated, which requires activation of excitatory synaptic input from the cortex, firing of the postsynaptic striatal neuron, and release of dopamine. This conjunction of activity related to sensory inputs, action, and dopamine results in long-lasting strengthening of corticostriatal synaptic connections (Wickens et al., Reference Wickens, Begg and Arbuthnott1996). Behaviorally reinforcing electrical stimulation of the brain also causes dopamine-dependent strengthening of corticostriatal synaptic connections (Reynolds et al., Reference Reynolds, Hyland and Wickens2001).

1.4 Relating the synaptic eligibility trace and the behavioral delay of reinforcement gradient

Electrophysiological experiments have shown that at the cellular level, there is a narrow time window of a few seconds during which dopamine is effective at strengthening synaptic connections (Cassenaer & Laurent, Reference Cassenaer and Laurent2012; He et al., Reference He, Huertas, Hong, Tie, Hell, Shouval and Kirkwood2015; Shindou et al., Reference Shindou, Shindou, Watanabe and Wickens2019; Yagishita et al., Reference Yagishita, Hayashi-Takagi, Ellis-Davies, Urakubo, Ishii and Kasai2014). If dopamine is released a few seconds too early, or too late, it is ineffective in promoting strengthening of neural connections. We refer to this time window as a synaptic eligibility trace.

At the behavioral level, there is also a restricted time window during which positive reinforcement can affect response learning (Renner, Reference Renner1964; Tarpy & Sawabini, Reference Tarpy and Sawabini1974). Reinforcement is more effective at shorter delays than at longer delays, and a delay reduces the effect of positive reinforcement on learning (Critchfield & Lattal, Reference Critchfield and Lattal1993; Dickinson, Watt, & Griffiths, Reference Dickinson, Watt and Griffiths1992; Lattal & Gleeson, Reference Lattal and Gleeson1990). This relation between delay and effectiveness of reinforcement is known as the delay of reinforcement gradient (Tarpy & Sawabini, Reference Tarpy and Sawabini1974). With natural rewards and no additional cues, the delay of reinforcement gradient has a half-life on the order of tens of seconds (Perin, Reference Perin1943).

In contrast to the delay of reinforcement gradient, the synaptic eligibility trace has a much shorter time course of a few seconds (Shindou et al., Reference Shindou, Shindou, Watanabe and Wickens2019). However, under conditions of delayed reinforcement, a sensory cue that reliably precedes reward can facilitate learning by anticipatory release of dopamine and thus reduce the effects of the delay of actual reward (Renner, Reference Renner1964; Tarpy & Sawabini, Reference Tarpy and Sawabini1974). Even when there is not an obvious sensory cue, the organism’s response, or the delay itself, can act as a predictive cue (Ferster, Reference Ferster1953; Garrud, Goodall, & Mackintosh, Reference Garrud, Goodall and Mackintosh1981; Winstanley, Theobald, Cardinal, & Robbins, Reference Winstanley, Theobald, Cardinal and Robbins2004). The predictive cue “bridges” the delay (Cardinal, Winstanley, Robbins, & Everitt, Reference Cardinal, Winstanley, Robbins and Everitt2004; Grice, Reference Grice1948) and, thus, there is timely release of dopamine at the cellular level. If this bridging mechanism is disabled by removing all predictive cue signals, for example, by using direct stimulation of brain dopamine neurons as the reinforcer, the effects of reinforcement are reduced by delays of as little as one second (Black, Belluzzi, & Stein, Reference Black, Belluzzi and Stein1985). Thus, it seems that the critical timing requirement for dopamine-dependent plasticity at the cellular level can be met, even when there is a more prolonged delay of reinforcement at the behavioral level, by the transfer of dopamine release from the reward to the reward-predicting cues.

1.5 The dopamine transfer deficit hypothesis

We have suggested above that transfer of the dopamine signal from reward to reward-predicting cue ensures that dopamine release occurs at the right time to strengthen synaptic connections at the cellular level, even when the behavioral reinforcer is delayed. However, the success of this depends on the ability to learn the cue-reward association and complete the transfer of the dopamine signal to the cue. We have previously considered the possible consequences of failure to learn the cue-reward association. We refer to this as the dopamine transfer deficit (DTD) hypothesis (Tripp & Wickens, Reference Tripp and Wickens2008, Reference Tripp and Wickens2009), as illustrated in Fig. 1b. Other authors have also proposed that reduced dopamine functioning causes altered processing of reward in individuals with ADHD (Levy, Reference Levy1991; Sagvolden, Johansen, Aase, & Russell, Reference Sagvolden, Johansen, Aase and Russell2005) in ADHD. However, DTD is unique in focusing specifically on the timing of the phasic dopamine response.

In developing the DTD hypothesis, we tried to predict the behavioral characteristics that would result from deficient transfer of the dopamine response from rewards to cues. In a condition where dopamine release did not develop in response to a reward-predicting cue, the dopamine signal at the cellular level would be delayed until the actual reward occurred. Under such conditions, dopamine-dependent strengthening of connections would be reduced or not occur at all. Several predictions follow:

1.6 Delay of reinforcement gradient

In the absence of dopamine release by reward-predicting cues, even very short delays of a few seconds would reduce the reinforcing effect of the delayed reinforcer. Thus, deficient transfer of the dopamine response from rewards to cues is predicted to cause a steeper delay of reinforcement gradient.

1.7 Partial reinforcement effects

Under conditions of partial reinforcement – a schedule in which not every response is reinforced – dopamine release by cues that are present on every trial would normally provide continuous reinforcement at the cellular level. In the absence of such dopamine release, learning under partial reinforcement will be slower. Furthermore, it is well established that although acquisition of learning is slower under partial reinforcement, the learning that is acquired is more resistant to extinction. This is called the partial reinforcement extinction effect (Myers, Reference Myers1960). The DTD hypothesis predicts slower learning under partial reinforcement and faster extinction (less behavioral persistence) of learned behavior because of reduced dopamine response to reward-predicting cues when reinforcement is stopped (Tripp & Wickens, Reference Tripp and Wickens2008).

1.8 Integration of reinforcement history

Behavior is not only controlled by individual instances of reinforcement but also by an internal representation of the integrated reinforcement history (Killeen & Sitomer, Reference Killeen and Sitomer2003; Okouchi & Lattal, Reference Okouchi and Lattal2006), so that responses that typically result in reinforcement are selected over those that most recently resulted in reinforcement (Tripp & Alsop, Reference Tripp and Alsop1999). Failure of dopamine transfer would cause increased sensitivity to individual instances of reinforcement.

In light of these predictions in the following sections, we review behavioral studies on the spontaneously hypertensive rat (SHR) model for ADHD, and humans with ADHD, before considering the implications of variations in dopamine transfer for symptoms and personality dimensions of individuals with ADHD and typically developing individuals.

3.1 Steeper delay of reinforcement gradient in rodent models and ADHD

Evaluating an animal model for ADHD requires careful attention to the symptoms of the human disorder. One way to ensure a connection between non-human animal and human studies is to use the same task for both. In some cases, nearly identical tasks have been used in studies of SHRs and humans with ADHD. These include a rat task adapted for use in children, and conversely, a human task adapted for use in rats.

An example of a rat task adapted for children is a fixed-interval (FI) schedule that has been extensively used in the investigation of behavioral characteristics of the SHR (Sagvolden et al., Reference Sagvolden, Pettersen and Larsen1993). In that schedule, rewards are delivered for the first response emitted after a fixed period of time has elapsed, but not for responses emitted earlier. The FI is then restarted for a number of repetitions. In such schedules, there is usually a decrease in responses after the reward, followed by an increase in responses toward the end of the FI.

The characteristic pattern of FI responding has been interpreted as a reflection of the delay of reinforcement gradient. The basis for this interpretation is that a reinforcer not only increases the probability of the response that produced it but to a lesser extent also increases the probability of earlier responses (Catania, Reference Catania1971; Johansen et al., Reference Johansen, Killeen, Russell, Tripp, Wickens, Tannock and Sagvolden2009; Killeen, Reference Killeen2011). This is because, in the FI schedule, the earlier responses in the interval are separated from the reinforcer delivery by a longer delay than the responses occurring later. Assuming that the effects of reinforcement are distributed according to a delay of reinforcement gradient, the earlier responses will be strengthened less than the later responses. Thus, after learning, the earlier responses occur at a lower rate than the later responses (Catania, Sagvolden, & Keller, Reference Catania, Sagvolden and Keller1988; Wearden & Lejeune, Reference Wearden and Lejeune2006). An example is shown in Fig. 2b. The increase in responding over the FI is, therefore, treated as an indirect measure of the delay of reinforcement gradient.

Figure 2. Delay of reinforcement gradient interpretation of fixed-interval responding. (a) Classic representation of the control exerted by delayed reinforcers. Habit strength is plotted as a function of time of reinforcement, based on the formula ${1 \over L} = {0.22^{ - 0.215t}} - 0.0188t + 0.32 $ fitted to experimentally measured latency or responses, L, for reinforcement at different delay times, t, (Perin, Reference Perin1943). (b) Example of FI responding showing the increase in response rate as the time of reinforcer delivery approaches for four different strains of rat. GH, Genetically Hypertensive, SHR, Spontaneously Hypertensive, WKY, Wistar Kiyoto, WI, Wistar. Redrawn from Wickens et al. (Reference Wickens, Macfarlane, Booker and McNaughton2004).

On the FI schedule, SHRs produce more responses than comparison strains, with responses increasing more steeply over the FI (Sagvolden, Hendley et al., Reference Sagvolden, Hendley and Knardahl1992; Sagvolden et al., Reference Sagvolden, Pettersen and Larsen1993). This finding is robust and has been replicated in other laboratories (Orduna, Reference Orduna2015; Wickens, Macfarlane, Booker, & McNaughton, Reference Wickens, Macfarlane, Booker and McNaughton2004). Thus, their FI behavior indicates that the SHRs have a greater sensitivity to delay of reinforcement than comparison strains.

On a similar FI schedule, children with ADHD made significantly more responses during the FI than typically developing children, with increased responses toward the end of the FI (Sagvolden et al., Reference Sagvolden, Aase, Zeiner and Berger1998). As in the SHR, this was interpreted as indication of a steeper-than-normal delay of reinforcement gradient in children with ADHD, operating over a similar timescale of seconds (Sagvolden et al., Reference Sagvolden, Aase, Zeiner and Berger1998).

3.2 Increased preference for immediate reinforcement in rodent models and ADHD

The converse example of a human task that was subsequently adapted for rats is a signal-detection task developed for studying sensitivity to delay of reinforcement in children with ADHD (Tripp & Alsop, Reference Tripp and Alsop2001). In this task, children had to press one of two buttons to identify one of two similar stimuli. Correct identification of one stimulus was immediately reinforced, followed by a 3.5 s delay before the start of the next trial. The other stimulus was reinforced after a 3.5 s delay. Both typically developing children and children with ADHD showed a bias toward immediate reinforcement, but the bias in the ADHD group was greater than in controls. These results showed that in this task, children with ADHD are not more sensitive to delay itself, as suggested by the delay aversion hypothesis (Sonuga-Barke, Taylor, Sembi, & Smith, Reference Sonuga-Barke, Taylor, Sembi and Smith1992), but specifically have higher sensitivity to delay of reinforcement.

In the animal version of the task, rats were rewarded for lever pressing one of two available levers on each trial (Sutherland et al., Reference Sutherland, Alsop, McNaughton, Hyland, Tripp and Wickens2009). One lever was associated with immediate delivery of a food pellet, and the other lever with delivery of a food pellet after a 10-s delay. Compared to their control strain, SHRs showed a greater bias toward immediate reinforcement. These results show that, like children with ADHD, SHRs have higher sensitivity to delay of reinforcement than comparison strains.

Surprisingly, genetically hypertensive (GH) rats showed similarly greater bias toward immediate reinforcement than the comparison strain. The GH rat was derived independently of the SHR by a similar process of selective breeding for high blood pressure (Simpson et al., Reference Simpson, Phelan, Clark, Jones, Gresson, Lee and Bird1973; Smirk & Hall, Reference Smirk and Hall1958). Like the SHR (Hendley et al., Reference Hendley, Atwater, Myers and Whitehorn1983), the genetically determined behavioral characteristics of the GH rat are dissociated from high blood pressure (Wickens et al., Reference Wickens, Macfarlane, Booker and McNaughton2004). Nonetheless, the GH rat shows FI responding (Wickens et al., Reference Wickens, Macfarlane, Booker and McNaughton2004) and sensitivity to delay of reinforcement (Sutherland et al., Reference Sutherland, Alsop, McNaughton, Hyland, Tripp and Wickens2009) similar to the SHR. In addition, the GH rat also showed a tendency to be more influenced by individual instances of reinforcement (Sutherland et al., Reference Sutherland, Alsop, McNaughton, Hyland, Tripp and Wickens2009).

3.3 Steeper delay discounting in rodent models and ADHD

Individuals with ADHD tend to choose small immediate rewards over larger delayed rewards in both choice delay and temporal discounting paradigms (Antrop et al., Reference Antrop, Stock, Verte, Wiersema, Baeyens and Roeyers2006; Firestone & Douglas, Reference Firestone and Douglas1975; Hoerger & Mace, Reference Hoerger and Mace2006; Jackson & MacKillop, Reference Jackson and MacKillop2016; Marx, Hacker, Yu, Cortese, & Sonuga-Barke, Reference Marx, Hacker, Yu, Cortese and Sonuga-Barke2021; Patros et al., Reference Patros, Alderson, Kasper, Tarle, Lea and Hudec2016; Scheres et al., Reference Scheres, Dijkstra, Ainslie, Balkan, Reynolds, Sonuga-Barke and Castellanos2006; Sonuga-Barke et al., Reference Sonuga-Barke, Taylor, Sembi and Smith1992). Although this is also true of typically developing individuals, the tendency is greater in individuals with ADHD. Most human studies have used temporal discounting tasks, which are also known as intertemporal choice or delay discounting tasks (Marx et al., Reference Marx, Hacker, Yu, Cortese and Sonuga-Barke2021; Sonuga-Barke, Sergeant, Nigg, & Willcutt, Reference Sonuga-Barke, Sergeant, Nigg and Willcutt2008; Sonuga-Barke et al., Reference Sonuga-Barke, Taylor, Sembi and Smith1992). These tasks measure the tendency to place less value on rewards that are delayed, often modeled as a discount curve indicating the subjective value of a reward as a function of delay to its receipt.

Although often lumped together with delay of reinforcement, there are fundamental differences in delay of reinforcement and measures of temporal discounting, both in concept and in task design. In delay of reinforcement tasks, the subject (whether human or animal) receives a reinforcer after a delay and the effect of the delay on future responses is measured. In contrast, rewards and delays in human temporal discounting tasks are usually hypotheticals and communicated verbally: they have never and never will be experienced by the individual (Killeen, Reference Killeen2011). Presumably, the decision to delay discounting tasks is based on experience of actual delay of reinforcement in the past, combined with a prediction of how the hypothetical reward will feel in the future after an imagined delay. For example, in such tasks, the participant is given a series of choices between small immediate and larger delayed rewards (such as $50 now or $100 in a year). In the rodent tasks, rats make choices among food rewards that they have been trained to associate with different delays.

Comparison of delay discounting measures between rat strains has yielded equivocal results. Some researchers have reported steeper discounting rates in the SHRs (Aparicio et al., Reference Aparicio, Hennigan, Mulligan and Alonso-Alvarez2019; Bizot et al., Reference Bizot, Chenault, Houze, Herpin, David, Pothion and Trovero2007; Carbajal et al., Reference Carbajal, Bounmy, Harrison, Nolen, Regan, Williams and Sable2023; Fox et al., Reference Fox, Hand and Reilly2008; Orduna, Reference Orduna2015). Others have reported no difference between SHR and comparison strains (Adriani et al., Reference Adriani, Caprioli, Granstrem, Carli and Laviola2003; Garcia & Kirkpatrick, Reference Garcia and Kirkpatrick2013; Gonzalez-Barriga & Orduna, Reference Gonzalez-Barriga and Orduna2022; Ramos, Lopez-Tolsa, Sjoberg, & Pellon, Reference Ramos, Lopez-Tolsa, Sjoberg and Pellon2019).

It is unlikely that rodents decide based on rather than imagined and discounted future reward, raising questions about whether the rodent tasks measure the same constructs as human tasks (Hayden, Reference Hayden2016). The responses of rodents (and possibly humans) may, for example, indicate the effect of conditioned reinforcers established during past experiences of delayed outcomes (Killeen, Reference Killeen2011; Smith, Southern, & Kirkpatrick, Reference Smith, Southern and Kirkpatrick2023). Humans and rodents show similarly shaped discounting curves, but the discounting rate in humans is several orders of magnitude slower (months) than that measured in animals (seconds) (Hayden, Reference Hayden2016). Nevertheless, the study of delay discounting in animals has provided useful information about the neural substrates involved (Fobbs & Mizumori, Reference Fobbs and Mizumori2017).

3.4 DTD and symptoms of ADHD

The core behavioral differences predicted by the DTD hypothesis are described in Section 1. They include a steeper delay of reinforcement gradient, slower learning under partial reinforcement, faster extinction of learned behavior, a reduced partial reinforcement extinction effect (as recently demonstrated; Hulsbosch et al., Reference Hulsbosch, Beckers, De Meyer, Danckaerts, Van Liefferinge, Tripp and Van der Oord2023), and increased sensitivity to individual instances of reinforcement. From these core behavioral differences – which are also present in the SHR – we have previously proposed that DTD would cause impulsivity and inattention symptoms described in the diagnostic criteria for ADHD in DSM and ICD systems (Tripp & Wickens, Reference Tripp and Wickens2008).

The symptoms predicted by the DTD hypothesis include the following: Greater control of behavior by immediate rather than delayed reinforcement would lead to less on-task behavior in situations of infrequent reinforcement. In the absence of constant supervision, unscheduled reinforcing events would control behavior. For example, a child in the classroom may find more reinforcement in looking out the window than attending to schoolwork, for which reward is delayed and infrequent. This would produce symptoms of inattention such as failing to finish tasks and reluctance to engage in tasks requiring sustained effort in the absence of reinforcement. In situations of delayed reinforcement, DTD would also lead to impulsive behaviors such as difficulty awaiting turns or intruding on others because of the immediate reinforcement of such actions.

3.5 Personality dimensions relevant to the DTD hypothesis

As clinicians and neurobiologists, we are inclined to view personality as an effect of psychological processes on behavior, based on underlying neurobiological mechanisms. In this view, the correlational structure detected in personality research may reflect the “common cause” of variations in underlying neural mechanisms. However, the empirical basis for the structure of most personality models is self-reporting and analysis of correlations among items, rather than behavioral tests or putative neural mechanisms. Although personality traits can be said to “predict” behavioral characteristics in a statistical sense, such correlations do not establish the direction of causality. Here we suggest that variations in the transfer of phasic dopamine signals from reward to predictive cues, through its effect on core processes related to timing of behavioral reinforcement, is a potential driver of dimensions that are missing or understated in current personality models.

If ADHD is viewed as an exaggeration of normal personality traits, lying at different points on the same continua as typically developing individuals (the “spectrum” model), then we might expect the relevant personality dimensions and ADHD symptoms to share a common neurobiological basis. However, a meta-analysis examining personality dimensions of the Five-Factor Model (which defines five dimensions: neuroticism, agreeableness, conscientiousness, extraversion and openness) in relation to ADHD symptoms concluded that although there was some support for the spectrum model, “the shared variance for all significant relations between personality and ADHD was never more than 50% suggesting that the spectrum model alone does not provide sufficient explanation for the association between personality and ADHD” (Gomez & Corr, Reference Gomez and Corr2014). We would argue that this may indicate that one or more relevant dimensions are missing from the personality model. The DTD hypothesis suggests some facets of those dimensions.

Although the DTD hypothesis (Tripp & Wickens, Reference Tripp and Wickens2008) was originally developed from considering the implications of a putative “deficiency” of transfer of dopamine responses from reward to reward predictive cues, the degree of transfer is likely to vary between individuals. Such variations in association of cues with reward would lead to corresponding variations in specific behavioral characteristics. For example, variation in dopamine response to reward-predicting cues would affect the degree of perseverance in pursuit of goals in the absence of continuous reinforcement. Similarly, variation in the integration of reinforcement history would lead to a tendency to be distracted by individual instances of reward. Also, variations in the delay of reinforcement gradient would lead to different degrees of impulsive choice. Thus, the DTD hypothesis predicts variations in impulsivity, distractibility, and persistence, according to individual differences in the degree of dopamine transfer. These behavioral characteristics do not exactly align with the personality dimensions derived from the Five-Factor model, although there is some overlap, as we discuss in the following section.

3.6 Impulsivity, delay of reinforcement gradients, and temporal discounting

Impulsivity is itself a broad concept that includes multiple dimensions. For example, Whiteside and Lynam (Reference Whiteside and Lynam2003) proposed that impulsivity includes sensation seeking, lack of premeditation, lack of perseverance, negative urgency, and positive urgency. Alternatively, MacKillop et al. (Reference MacKillop, Weafer, Gray, Oshri, Palmer and de Wit2016) suggest that the latent structure among multiple measures of impulsivity has three broad categories, namely impulsive choice; impulsive action; and impulsive personality traits, reflecting self-reported attributions of self-regulatory capacity. Such self-report scales focus more on the tendency to act without forethought and inability to inhibit responses, rather sensitivity to delay of reward. However, self-report measures of impulsivity personality traits have weak relations to delay discounting (Bobova, Finn, Rickert, & Lucas, Reference Bobova, Finn, Rickert and Lucas2009; Janis & Nock, Reference Janis and Nock2009; Odum, Reference Odum2011a). Therefore, we suggest that the facets of impulsivity, most relevant to the DTD hypothesis, are impulsive choice and lack of perseverance.

Initial studies of the relationship between delay discounting and personality suggested that delay discounting was related to neuroticism and conscientiousness (Mahalingam, Stillwell, Kosinski, Rust, & Kogan, Reference Mahalingam, Stillwell, Kosinski, Rust and Kogan2014; Manning et al., Reference Manning, Hedden, Wickens, Whitfield-Gabrieli, Prelec and Gabrieli2014). However, in a large sample, Yeh et al. (Reference Yeh, Myerson and Green2021) found that discounting was not correlated with neuroticism or conscientiousness scales, and suggested that delay discounting is an important individual difference characteristic in its own right. Thus, the delay discounting rate may be related to an additional personality dimension that is not included In the Big Five, namely impulsivity (Odum, Reference Odum2011b). Consistent with this, delay discounting measures are proving useful in understanding links between neural systems and behavior in healthy individuals as well as understanding psychopathology (Lempert, Steinglass, Pinto, Kable, & Simpson, Reference Lempert, Steinglass, Pinto, Kable and Simpson2019).

3.7 Persistence

Resistance to extinction – in other words, persistent effort in the absence of reinforcement – is likely to depend on dopamine transfer. As mentioned above, the DTD hypothesis predicted a reduced partial reinforcement extinction effect in children with ADHD, which was recently supported by an experimental study (Hulsbosch et al., Reference Hulsbosch, Beckers, De Meyer, Danckaerts, Van Liefferinge, Tripp and Van der Oord2023). In a meta-analytic review of ADHD and personality measures, inattention was associated with a lack of perseverance (Gomez, Stavropoulos, Watson, Brown, & Chen, Reference Gomez, Stavropoulos, Watson, Brown and Chen2022), and the association between ADHD and perseverance was moderated by age (stronger in children than adults) and source (stronger in clinical samples than community samples). Thus, persistence appears likely to be a facet of personality dimensions related to variations in dopamine transfer.

3.8 Other personality dimensions

There are some suggestions that extraversion may be related to functional variation in dopamine responses that attach incentive value to reward-predicting cues. For example, Depue and Collins (Reference Depue and Collins1999) argue that variation in encoding of incentive salience – the intensity of stimulus representations that have become associated with reward through experience – is the main source of individual differences in extraversion. Consistent with this hypothesis studies measuring reward prediction (Smillie, Cooper, & Pickering, Reference Smillie, Cooper and Pickering2011) or reward sensitivity (Blain, Sassenberg, Xi, Zhao, & DeYoung, Reference Blain, Sassenberg, Xi, Zhao and DeYoung2021) have shown positive associations with measures of extraversion. More extraverted individuals show a greater preference for immediate rewards (Hirsh, Morisano, & Peterson, Reference Hirsh, Morisano and Peterson2008; Ostaszewski, Reference Ostaszewski1996). However, although an association of ADHD with higher extraversion has been found in some studies (Gomez & Corr, Reference Gomez and Corr2014), the finding is inconsistent across studies and the association may be limited to self-reports (Nigg et al., Reference Nigg, John, Blaskey, Huang-Pollock, Willcutt, Hinshaw and Pennington2002).

In relation to the remaining personality dimensions, Gomez and Corr (Reference Gomez and Corr2014) concluded, on the basis of a meta-analysis of 40 data sets, that ADHD symptoms of inattention and hyperactivity/impulsivity were associated with measures of conscientiousness, agreeableness, and neuroticism. Consistent findings have been reported from more recent analyses (Jacobsson, Hopwood, Soderpalm, & Nilsson, Reference Jacobsson, Hopwood, Soderpalm and Nilsson2021). There is no obvious relationship between these personality dimensions and variations in dopamine transfer, although conscientiousness might tap into persistence and agreeableness into distractibility.