Imagine you are in an orchard, trying to decide which of the many apples to pick. On what do you base your decision? Most research into this type of decision-making has focused on how the brain uses visual information – about features such as colour, size and shape – to make a choice. But what about the effort required to obtain the apple? Does an apple at the top of the tree look more or less tempting than the low-hanging fruit?

To answer this kind of question, Hagura et al. asked volunteers to decide whether dots on a screen were moving to the left or to the right. The volunteers indicated their choice by moving one of two levers. If they thought the dots were moving to the right, they moved a lever in their right hand. If they thought the dots were moving to the left, they moved a lever in their left hand. What the volunteers did not know, however, is that one of the levers was slightly heavier and therefore harder to move than the other.

Hagura et al. found that the volunteers biased their decisions away from the direction that would require the most effort. If the right-hand lever was heavier, the volunteers decided that dots with ambiguous motion were moving to the left. Those for whom the left-hand lever was heavier felt that the same dots were moving to the right. The participants showed this bias despite failing to notice that the levers had different weights. Moreover, they continued to show the bias even when subsequently asked to simply say their answers rather than use the levers.

These results indicate that the effort required to act on a decision can influence the decision itself. The fact that participants were biased even when responding verbally, and despite being unaware that the levers differed in weight, suggests that they were not deliberately choosing the easier option. Instead, the cost to act changed how they perceived the stimuli themselves. The findings also suggest that it might be possible to help people make better decisions by designing environments in which less favourable options require more effort.

In laboratory experiments, participants are often asked to make decisions that are purely based on the features of the sensory input – a process that we refer to here as perceptual decision-making. However, in many of our daily situations, decisions are made in a behavioural context, in which the action that follow our decisions can differ dramatically in terms of required physical effort (or the motor cost). For example, in the orchard, one may aim to pick the reddest-looking apple from the tree. Some of the apples may be hanging high-up on the tree, which will require more effort to pick compared to other fruits hanging on the lower branch. In such situations, does the difference in the motor cost between the options influence the decision of which fruit to pick? If so, is such influence a result of serial integration between the perceptual decision (i.e. decision based on the visual feature) and the motor decision (i.e. decision for action selection to avoid the effortful action) at the output stage, or is the perceptual decision itself is affected by the cost on the downstream action? It has been shown that physical effort is used in motor planning (Huang et al., 2012; Izawa et al., 2008), and the physical effort to obtain a reward can influence behavioural decisions (Prévost et al., 2010; Hosokawa et al., 2013). Moreover, the uncertainty in perceptual decisions is transmitted to the motor system, influencing the parameters of action control (de Lange et al., 2013). However, it remains unclear whether the motor cost is simply integrated with the perceptual decision to optimise the expected utility (Burk et al., 2014; Cos et al., 2014), or whether the preceding experience of unequal motor costs can recursively influence the perceptual decision itself. Here, we show that manipulating the motor response cost for arm movements during a visual motion discrimination task changes not only the decision when responding with the arm, but also when reporting the perceptual decisions verbally.

First, in Experiment 1, we examined if the decision of the visual motion direction can be biased when one of the two responses requires more effort. Ten right-handed participants observed a moving random-dot stimulus and made decisions about the direction of motion (leftward or rightward) (Britten et al., 1992). Participants held two robotic manipulanda, one in each hand. They indicated their decision by either moving their left hand (indicating leftward decision) or right hand (rightward decision, Figure 1A). In the baseline phase, the resistance for moving the manipulanda was the same for both hands (velocity-dependent resistance: 0.10 Ns/cm). In the subsequent induction phase, the resistance for the left hand increased by a small amount each time the participant moved the left hand (0.0008 Ns/cm; Figure 1B). Because the change was gradual, most of the participants did not report becoming aware of the increased motor cost when asked afterwards, even though their left hand was eventually exposed to 1.8 times greater resistance than the right (0.18 Ns/cm for left, 0.10 Ns/cm for right; see Materials and methods and Figure 1—figure supplement 1). This procedure was employed to minimise any cognitive strategy participants may use, such as explicitly avoiding the costly hand response regardless of the decision about the visual stimulus. In the test phase, participant then continued to perform the visual discrimination task under the accumulated asymmetry in manual resistance. We plotted the proportion of rightward judgement against different stimulus intensities, and determined the point of subjective equality (PSE, the point at which participants judge 50% of the trials to go rightward) for both the baseline and the test phase (Figure 1D). If the increased physical resistance for expressing leftward judgements was incorporated into the decision, the proportion of ‘leftward’ judgements should decrease in the test phase compared to the baseline, resulting in the shift of PSE towards the left (Figure 1D). As expected, the PSE shifted towards the left from baseline to test phase (−4.33%, paired t-test (2-tailed): t₉ = 2.43, p=0.038, d = 0.76, 8/10 individuals showed the effect) (Figure 1D–E, Figure 1—source data 1). This indicates that the participants started to avoid making motion direction decisions in which the response is costly.

Figure 1 with 1 supplement see all

Download asset Open asset

Figure 1—source data 1

Individual PSE shift for Experiments 1–4, which is the data summarised in Figure 1E.

https://doi.org/10.7554/eLife.18422.004

Download elife-18422-fig1-data1-v2.xlsx

In Experiment 2, we examined whether motor cost and visual features need to be directly associated, or whether simply gaining experience of one action being more effortful than the other is sufficient to bias subsequent decisions. The baseline and the test phase involved judging direction of visual dot motion, as in Experiment 1. However, the induction phase was now replaced with a simple reaching movement, in which the participants moved their left or right hand according to a simple leftward or rightward arrow presented in the centre of the screen. As in Experiment 1, the resistance for moving the left hand was gradually increased. The motor cost during the induction phase was not associated with any motion direction judgement; the participants were only exposed to the gradually increasing motor cost differences between the two hands. Again, PSE significantly shifted leftwards in the test phase (Figure 1E, Figure 1—source data 1; baseline vs. test; mean −4.26%, paired t-test (two-tailed); t₈ = 3.91, p=0.005, d = 1.3, 8/9 individuals showed the effect). This indicates that the direct association of higher motor cost with a specific decision during the induction phase is not critical for inducing the bias. This may suggest that the (implicit) knowledge about the response costs is sufficient to recursively influence the decision. Alternatively, these results could indicate that the bias is only transiently induced during the test phase itself.

In Experiments 1 and 2, we showed that manual motor costs reliably bias decisions that involve these manual response. Where in the process of translating a stimulus into a response does this bias arise? A simple model posits that decision-making occurs in three sequential stages (Gold and Ding, 2013). First, features of the sensory input are extracted and encoded as a sensory representation. Second, a categorical decision is made based on this sensory representation (decision layer). Third, the output from the decision layer is transmitted to the relevant effector for the response (Figure 2). One possibility is that the motor cost only biases the decision layer in the context of the specific response (Oliveira et al., 2010) (Figure 2A). In other words, the motor cost only influences decisions when the participant anticipates to perform the action associated with the motor cost, thus, the decision simply takes into account the upcoming motor costs. Alternatively, the repeated exposure to the manual motor cost may affect the perceptual decision about this type of stimulus in general, no matter which effector is used to make a response (Figure 2B) (Bennur and Gold, 2011; Filimon et al., 2013). Finally, the motor cost could also directly bias the sensory representation (Figure 2C), affecting the initial encoding of the information before it is transmitted to the decision layer. Only in the two latter scenarios, should the bias observed during the manual decisions generalise to decisions expressed with a different effector. In Experiment 3, we therefore examined whether a hand-specific motor cost could also influence a visual judgement that used a vocal response. A manual to vocal transfer of the motor cost effect would indicate that the motor cost influences the decision about the visual stimulus itself (i.e. perceptual decision), not the decision coupled with the effector selection (i.e. motor decision).

Figure 2

Download asset Open asset

Fourteen new participants performed the visual motion discrimination as in Experiment 1. In the induction phase, we gradually increased the resistance for one of the hands while participants performed manual decisions as in Experiment 1. The resistance was increased for half of the participants (7) on the left hand, and for the other half on the right hand, accounting for any hand-dependent effects. To analyse these left and right resistance increase data together, we aligned the data depending on the side of the resistance applied by assigning negative motion coherence level to the motion direction associated with the direction of the resistance. During the manual task, participants moved their left or right hand according to their perceived motion direction. For the vocal task, participants indicated the direction of the motion by vocally responding ‘left’ or ‘right’ (Figure 3—figure supplement 1A) without moving their hands. During the baseline and the test phase, participants alternated between tasks: each 10 trials of manual judgements were followed by 10 trials of vocal judgements (Figure 1C). This ‘top-up’ procedure is commonly used to assess the effect of sensory adaptation on the subsequent perceptual judgements (Fujisaki et al., 2004). If the bias induced by the motor cost is affecting the decision regardless of the response effector, the vocal decision should be also biased towards the same direction as the manual decision.

For the manual task, the result of Experiment 1 was replicated. The exposure to the resistance made the PSE to shift significantly away from the stimulus associated with the costlier movement (baseline vs. test, mean: −7.06%, paired t-test:t₁₃ = 2.94, p=0.012, d = 0.78, 12/14 individuals showed the effect) (Figure 1E, Figure 1—source data 1). More importantly, for the vocal task, judgement also shifted to the same direction as the manual task (baseline vs. test, mean: −3.00%, paired t-test:t₁₃ = 2.44, p=0.030, d = 0.65, 12/14 individuals showed the effect) (Figure 1E, Figure 1—source data 1), even though the motor cost for the vocal responses was not manipulated. Since the direction of manual motor cost was counterbalanced across participants, this finding cannot be explained by any time-dependent drift of the decision towards one of the directions. This result suggests that the bias induced by the manual motor cost transfers to decisions expressed with other effectors.

Although in Experiment 3, the response effector differed between the manual and the vocal task, the abstract response code (‘left’/ ‘right’) remained the same between the two tasks. Therefore, it is possible that the manual motor cost got associated with these semantic labels, but did not necessarily influence the stimulus-based perceptual decision itself. To test this possibility, in Experiment 4, we again examined the manual-to-vocal transfer of effect caused by the motor cost, but this time varied not only the response effector, but also the response codes between the two tasks. Twelve new participants performed visual motion judgements, where in the baseline and the test phase, manual decisions and the vocal decisions alternated in a mini-block of 11 and 7 trials, respectively (Figure 3A). The induction phase involved only the manual task, with gradually increasing left hand resistance. As in Experiment 3, the manual task was a left-right motion discrimination task. The vocal task, however, was changed to the motion detection task. Participants were asked to detect a near threshold coherent motion by vocally responding ‘yes’ or ‘no’. The to-be-detected motion direction (left or right) was instructed at random before each trial (Figure 3—figure supplement 1B). Half of the trials included left or right coherent motion, and in the other half, the coherent motion was absent (0% coherence).

Figure 3 with 1 supplement see all

Download asset Open asset

Figure 3—source data 1

Individual criterion shift of the vocal trials (Experiment 4; summarised in Figure 3B), individual PSE shift for the vocal trials across different trials (Experiment 3; summarised in Figure 3C) and individual criterion shift for the vocal trials across different trials (Experiment 4; summarised in Figure 3D).

https://doi.org/10.7554/eLife.18422.008

Download elife-18422-fig3-data1-v2.xlsx

For the manual task, a significant shift of PSE was observed again, reflecting the avoidance of the costly decision (baseline vs. test, mean: −4.39%, paired t-test: t₁₁ = 2.88, p=0.015, d = 0.75, 9/12 individuals showed the effect) (Figure 1E, Figure 1—source data 1). For the vocal task, participants’ judgement criterion for leftward motion detection became more conservative after being exposed to the manual motor cost, which was not the case for the rightward motion. The interaction between the phase of the experiment (baseline/test) and the visual motion direction (left/right) was significant, (F_1,11 = 6.76, p=0.025, ${\eta }^2$=0.36) (Figure 3B, Figure 3—source data 1). Since the manual task required a left-right decision and the vocal task a yes-no decision, the abstract response code of these two task were different. Therefore, significant manual-to-vocal transfer cannot be simply explained by the motor cost inducing a bias for choosing a particular type of abstract response label. Instead, the results indicate that the motor cost influenced the perceptual decision – that is the decision based on the feature of the visual stimulus input – itself.

Together, these results demonstrate that the motor cost on the downstream response can recursively change how the input visual stimulus is transformed into the decision; at the level of sensory representation or at the decision layer. In contrast to the criterion, the sensitivity (d’) for the motion detection did not change for either visual motion direction (F_1,11 = 0.44, p=0.52,η²=0.04) (Figure 3—figure supplement 1C). This indicates that the motor cost did not increase or decrease the signal to noise ratio (gain) of the motion signal to one specific direction.

Until now, we have shown that the motor cost can bias the decision based on a visual stimulus independent from the response effector or abstract response code. Finally, using a model-based approach, we tried to elucidate the processing stage in which the motor cost could have influenced the decision. We analysed both reaction time and choice data of the manual tasks (Experiments 1, 2, 3 and 4; n = 45) under the framework of diffusion decision model (DDM) (Ratcliff and McKoon, 2008). The DDM postulates that a decision variable temporally accumulates sensory evidence in favour of one decision (by increasing its value) or in favour of the alternative decision (by decreasing its value). When the decision variable hits a certain threshold level (decision bound), the decision is made and the response is triggered (Ratcliff and McKoon, 2008; Palmer et al., 2005) (Figure 4—figure supplement 1A). Under this framework, we examined whether the source of the manual decision bias that transferred to the vocal decisions occurred in the sensory representation of the stimulus that is accumulated (Figure 2C; sensory representation) or in the decision bound that is used to make the decision (Figure 2B; decision layer) (Ratcliff and McKoon, 2008; Palmer et al., 2005; Hanks et al., 2006).

If the bias is introduced at the sensory representation stage, it would increase the input signal (the perceived motion coherence) in the easier direction. We exclude the possibility that the motor cost made the sensory representation of the preferred direction more accurate (increased gain of the signal only in one direction) as we did not observe the discrimination sensitivity change (JND: just noticeable difference) between the baseline and the test condition across different experiments (t₄₄ = 0.26, p=0.77). Rather assumed that the motor cost would shift evidence accumulation towards the easier direction (sensory evidence model; Figure 4—figure supplement 1B). With this bias, the decision variable would drift towards the preferred decision even in the absence of any coherent visual motion.

Alternatively, we considered the possibility that the motor cost changed the decision bounds (decision layer), that is, the amount of evidence required for each of the choices. This change can be parsimoniously modelled as shift of the starting point of the accumulation process, which will consequently change the distance from the starting point to each decision bound (starting point model; Figure 4—figure supplement 1C). The sensory evidence and starting point models predict qualitatively similar pattern of choice probabilities (i.e. bias towards the direction to avoid the motor cost), but different pattern of decision times for correct trials across different motion intensities (Methods, Figure 4—figure supplement 1B–C). Therefore, by comparing whether which of the two models explains our data better (Hanks et al., 2006; Ding and Gold, 2012) (see Materials and method), we may infer the source of the bias.

Additionally to the starting point model and the sensory evidence model, we also fitted a model that allowed for both shifts simultaneously (full model). This allows us to directly compare the effect of each parameter. Also, to check whether the starting point or the sensory evidence shift was necessary to explain the data in the first place, we also prepared a baseline model which we did not model the starting point and/or the evidence accumulation shift, but only modelled the difference in non-decision time (baseline model: see Material and methods). Note that, since we did not record the reaction time of the vocal decisions, this analysis was restricted to model the bias during the manual decisions.

First, we fit each model to the average group data and compared the BIC weights by converting the Bayesian Information Criterion (BIC) for each model (Wagenmakers and Farrell, 2004). We then repeated this process 10,000 times, each time drawing 45 participants from our sample with replacement to obtain an estimate of the reliability of our conclusion. The results (Figure 4C–D, Figure 4—source data 1, Table 1) clearly indicate that the starting point model explained the data substantially better than the other models.

Figure 4 with 2 supplements see all

Download asset Open asset

Figure 4—source data 1

Individual starting point shift in the test phase from the baseline phase during the manual trials, estimated from the full-model DDM (summarised in Figure 4A), individual evidence accumulation shift in the test phase from the baseline phase during the manual trials, estimated from the full-model DDM (summarised in Figure 4B), data points consisting the psychometric function estimated from the DDM (starting point model) using the group averaged data (summarised in Figure 4C) and data points consisting the chronometric function estimated from the DDM (starting point model) using the group averaged data (summarised in Figure 4D).

https://doi.org/10.7554/eLife.18422.012

Download elife-18422-fig4-data1-v2.xlsx

Second, we compared the model parameter of the full model fitted to each participants’ individual data. Consistent with superior fit of the starting point model, we found a significant shift of the starting point (median; 5.6%, signed rank test: z₄₄ = 2.50, p=0.01, d = 0.32; Figure 4A, Figure 4—source data 1), but no significant change in the evidence accumulation (median; 1.38%, signed rank test: z₄₄ = 1.15, p=0.25, d = 0.21; Figure 4B, Figure 4—source data 1). Therefore, our data suggest that the motor cost biased the decisions by changing the decision layer (starting point) that transforms the input signal into the decision.

The DDM also contains a parameter that captures the motor response time that is independent from the decision (non-decision time; see Material and methods). In the baseline phase, there was no significant difference between the non-decision time of the hands (462.8 ms vs. 468.8 ms, paired t-test: t₄₄ = 0.94, p=0.34, d = 0.08). However, in the test phase, although the non-decision time became shorter for both the hands without resistance (dTA: −7.69 ms) and the hand with resistance (dTB: −27.7 ms), the decrease was larger for the hand with resistance (paired t-test: t₄₄ = 2.69, p=0.01, d = 0.64). This finding resembles a previous report (Ding and Gold, 2012), which demonstrated that the stimulation of the caudate neuron with visual motion directional tuning biased monkey’s motion direction judgements towards the neuron’s tuned direction, but at the same time decreased the estimated non-decision time for the response (eye movement) to the opposite (non-stimulated) direction.

While the DDM models were only fit to the choice probabilities (psychometric function) and the RT function of the correct trials (chronometric function, see Material and methods), we also checked whether the models could predict the patterns of RTs on error trials. For this purpose, we simulated individual trials using the estimated group parameters based on the starting point and the sensory evidence models (Table 2). For the correct trials, both model simulations showed similar tendencies; the RT reduced for the non-costly motion stimulus compared to the costly stimulus (Figure 4—figure supplement 2A–B left panel). The pattern of the error trial RT differed between the two simulations. For the starting point model, error RTs were shorter for the costly motions (non-costly decision), whereas the pattern was opposite for the sensory evidence model (Figure 4—figure supplement 2A,B). This is because, for the former, the distance between the starting point and the non-costly decision bound decreases, whereas for the latter, the drift rate increases towards the error decision direction for the costly stimulus (i.e. non-costly decision) (Mulder et al., 2012). The pattern of RTs for the experimental data (Figure 4—figure supplement 2C) was qualitatively similar to that of the starting point model. Therefore, the general pattern of error RTs supported our claim that the motor cost induced a starting point shift.

We showed that participants incorporate the cost of the response into the perceptual decision, and flexibly changes the way of interpreting the sensory environment. To further investigate the temporal dynamics of these flexible changes, we examined how the induced bias developed over the course of the 10 (Experiment 3) or 7 (Experiment 4) trials of vocal decisions following a series of manual decision trials (see detail in the Materials and methods). In Experiment 3, on average, PSE shift was slightly stronger for the vocal trials that immediately followed the manual task (mean shift of PSE; −3.51% for the first five trials (initial point of Figure 3C), >−3% for the rest, Figure 3—source data 1), although this time dependence did not reach significance (F_{5, 65}=0.60, p=0.75). In Experiment 4, the strength of bias (criterion shift) significantly decayed depending on the number of trials from the manual task, showing a stronger bias in the first 4 out of 7 trials (initial two time points of the curve. Figure 3D, Figure 3—source data 1, F_4,44 = 2.70 p=0.042, η²=0.2). These results indicate that the biasing effect may be relatively short-lived. However, the time scale of the retention is comparable to common perceptual adaptations, such as motion aftereffects. Mather et al. (Mather et al., 2008) shows that on maximum, the motion aftereffect lasts for 10 ~ 15 s. A single trial of our task takes at least, 3 ~ 4 s, so our effect lasted for 9 ~ 16 s. Therefore, our results indicate that, in the absence of any further confirmatory evidence of asymmetric response costs for the decision, the brain readapts relatively quickly to the new situation, resembling other examples of spontaneous decay in perceptual and motor adaptation phenomena (Mather et al., 2008; Webster, 2011; Smith et al., 2006). This shows that while perceptual decisions can be updated relatively quickly and flexibly, they do exhibit a substantial memory of past motor costs.

In this study, we showed that visual motion direction decisions can be biased by the cost of the action that is used to report the decision (the motor cost). Moreover, we demonstrated that the motor cost indeed affects the decision about the input stimulus identity, and not only the decision about which action to select.

Previous behavioural studies have shown that the perceptual decisions can be biased by changing the frequency (expectation) of stimulus presentation, or by manipulating the response-reward association for the correct/incorrect decisions (Mulder et al., 2012; Whiteley and Sahani, 2008). Here, we demonstrate that motor costs associated with the response can also bias the perceptual decisions, even when the response is made with a completely different effector that is not associated with increased motor cost (here, verbal instead of manual responses). Therefore, our study provides evidence that the cost on the response for perceptual decisions, which has been regarded as downstream output channel of the decision, can recursively influence the decision of the input stimulus itself. In other words, the observed bias not only reflects the serial integration of the transient motor cost into the ongoing decision (Burk et al., 2014), but represents a more global change in the way of transforming the sensory input to the decision, by taking the prior experience of the motor cost into account (Makin et al., 2010).

Congruent with our results, a recent study has shown that asymmetric biomechanical costs induces a bias into the decisions, and that this bias cannot be explained by strategically choosing the easier option when perceptual uncertainty is high (Marcos et al., 2015). The critical contribution of our study is to show that this influence on the decision process is not limited to the judgements involving the asymmetric motor cost, but generalised to judgement using vocal responses without motor cost manipulation. The present study parsimoniously shows that the motor cost influence is not simply due to the bias of decision at the motor preparation/execution stage.

Our DDM analysis readily explains the effect of motor cost as the change in the required amount of input evidence (i.e. shifts of the starting point) for the decision. Even though only fitted to the reaction times for correct trials (Palmer et al., 2005), this model also correctly predicted the RT pattern for error trials. These findings are consistent with previous literature that shows that increasing the presentation frequency or amount of reward for one of the choice biases decision in a way that can be best modelled as a starting point shift (Mulder et al., 2012). Indeed, it has been suggested that shifting the starting point of accumulation process is the optimal solution to account for such contextual changes (Bogacz et al., 2006; Simen et al., 2009). It should be noted, however, that alternative models involving collapsing bounds may perform better in situations in which the stimulus strength varies randomly (Hanks et al., 2011; Tajima et al., 2016, but also see Hawkins et al., 2015).

Electrophysiological studies have shown that the electrical stimulation of parietal or basal ganglia neurons can bias perceptual decisions (Hanks et al., 2006; Ding and Gold, 2012). These effects were explained by shifts of the starting point in the DDM framework (or equivalently the decision bounds), thus the change in the decision layer of the perceptual task. Therefore, these brain regions are likely candidate neuronal substrates where the motor cost interacts with the sensory input to bias the perceptual decision. Neurons in the lateral intraparietal area (LIP) code the feature that is relevant for the visual decision, independent of response type (Bennur and Gold, 2011). Experiments 3 and 4 similarly suggested motor-induced changes for the perceptual decision is independent of the effector used for response. The subcortical network in the basal ganglia has been suggested to represent the cost of action or the ‘vigour’ of movement initiation (Mazzoni et al., 2007). Prolonged exposure to altered motor costs during perceptual decisions may similarly change the response properties of these areas, altering how the system judges the sensory evidence from the environment.

How does our current finding relate to the existing theories of perceptual decision making? One of the recent theories is the intentional framework (Shadlen et al., 2008). This framework posits that decisions and actions are tightly coupled, with each decision maker separately accumulating the sensory evidence until the threshold level for the specific action is reached. In this scenario, any decision bias induced by imposing a motor cost to a specific action would not transfer to a decision performed by a different action, as there is no explicit communication between the multiple decision makers. Thus, our results indicate that perceptual decisions are either made centrally by a high-order process that is common across different actions (Filimon et al., 2013; Sun and Landy, 2016), or at least that different local decision makers exhibit a certain degree of mutual dependency, such as a shared cost (value) of the input stimuli (external environment).

In conclusion, we demonstrate that the motor cost involved in responding to a visual classification task is integrated into the perceptual decision process. Our everyday perceptual decisions seem to be solely based on the incoming sensory input. They may be, however, influenced by the preceding history of physical cost of responding to such input. The cost of our own actions, learned through the life-long experience of interacting with the environment, may partly define how we make perceptual decisions of our surroundings.

1. 1. Bennur S
   2. Gold JI

   (2011) Distinct representations of a perceptual decision and the associated oculomotor plan in the monkey lateral intraparietal area

   Journal of Neuroscience 31:913–921.

   https://doi.org/10.1523/JNEUROSCI.4417-10.2011

   • PubMed
   • Google Scholar
2. 1. Bogacz R
   2. Brown E
   3. Moehlis J
   4. Holmes P
   5. Cohen JD

   (2006) The physics of optimal decision making: a formal analysis of models of performance in two-alternative forced-choice tasks

   Psychological Review 113:700–765.

   https://doi.org/10.1037/0033-295X.113.4.700

   • PubMed
   • Google Scholar
3. 1. Britten KH
   2. Shadlen MN
   3. Newsome WT
   4. Movshon JA

   (1992)

   The analysis of visual motion: a comparison of neuronal and psychophysical performance

   Journal of Neuroscience 12:4745–4765.

   • PubMed
   • Google Scholar
4. 1. Burk D
   2. Ingram JN
   3. Franklin DW
   4. Shadlen MN
   5. Wolpert DM

   (2014) Motor effort alters changes of mind in sensorimotor decision making

   PLoS One 9:e92681.

   https://doi.org/10.1371/journal.pone.0092681

   • PubMed
   • Google Scholar
5. 1. Cos I
   2. Duque J
   3. Cisek P

   (2014) Rapid prediction of biomechanical costs during action decisions

   Journal of Neurophysiology 112:1256–1266.

   https://doi.org/10.1152/jn.00147.2014

   • PubMed
   • Google Scholar
6. 1. de Lange FP
   2. Rahnev DA
   3. Donner TH
   4. Lau H

   (2013) Prestimulus oscillatory activity over motor cortex reflects perceptual expectations

   Journal of Neuroscience 33:1400–1410.

   https://doi.org/10.1523/JNEUROSCI.1094-12.2013

   • PubMed
   • Google Scholar
7. 1. Ding L
   2. Gold JI

   (2012) Separate, causal roles of the caudate in saccadic choice and execution in a perceptual decision task

   Neuron 75:865–874.

   https://doi.org/10.1016/j.neuron.2012.07.021

   • PubMed
   • Google Scholar
8. 1. Donchin O
   2. Francis JT
   3. Shadmehr R

   (2003)

   Quantifying generalization from trial-by-trial behavior of adaptive systems that learn with basis functions: theory and experiments in human motor control

   Journal of Neuroscience 23:9032–9045.

   • PubMed
   • Google Scholar
9. 1. Drugowitsch J
   2. Moreno-Bote R
   3. Churchland AK
   4. Shadlen MN
   5. Pouget A

   (2012) The cost of accumulating evidence in perceptual decision making

   Journal of Neuroscience 32:3612–3628.

   https://doi.org/10.1523/JNEUROSCI.4010-11.2012

   • PubMed
   • Google Scholar
10. 1. Efron B

    (1979) Bootstrap methods: another look at the Jackknife

    The Annals of Statistics 7:1–26.

    https://doi.org/10.1214/aos/1176344552

    • Google Scholar
11. 1. Estes WK
    2. Maddox WT

    (2005) Risks of drawing inferences about cognitive processes from model fits to individual versus average performance

    Psychonomic Bulletin & Review 12:403–408.

    https://doi.org/10.3758/BF03193784

    • PubMed
    • Google Scholar
12. 1. Filimon F
    2. Philiastides MG
    3. Nelson JD
    4. Kloosterman NA
    5. Heekeren HR

    (2013) How embodied is perceptual decision making? Evidence for separate processing of perceptual and motor decisions

    Journal of Neuroscience 33:2121–2136.

    https://doi.org/10.1523/JNEUROSCI.2334-12.2013

    • PubMed
    • Google Scholar
13. 1. Fujisaki W
    2. Shimojo S
    3. Kashino M
    4. Nishida S

    (2004) Recalibration of audiovisual simultaneity

    Nature Neuroscience 7:773–778.

    https://doi.org/10.1038/nn1268

    • PubMed
    • Google Scholar
14. 1. Gold JI
    2. Ding L

    (2013) How mechanisms of perceptual decision-making affect the psychometric function

    Progress in Neurobiology 103:98–114.

    https://doi.org/10.1016/j.pneurobio.2012.05.008

    • PubMed
    • Google Scholar
15. 1. Hagura N
    2. Kanai R
    3. Orgs G
    4. Haggard P

    (2012) Ready steady slow: action preparation slows the subjective passage of time

    Proceedings of the Royal Society B: Biological Sciences 279:4399–4406.

    https://doi.org/10.1098/rspb.2012.1339

    • Google Scholar
16. 1. Hanks TD
    2. Ditterich J
    3. Shadlen MN

    (2006) Microstimulation of macaque area LIP affects decision-making in a motion discrimination task

    Nature Neuroscience 9:682–689.

    https://doi.org/10.1038/nn1683

    • PubMed
    • Google Scholar
17. 1. Hanks TD
    2. Mazurek ME
    3. Kiani R
    4. Hopp E
    5. Shadlen MN

    (2011) Elapsed decision time affects the weighting of prior probability in a perceptual decision task

    Journal of Neuroscience 31:6339–6352.

    https://doi.org/10.1523/JNEUROSCI.5613-10.2011

    • PubMed
    • Google Scholar
18. 1. Hawkins GE
    2. Forstmann BU
    3. Wagenmakers EJ
    4. Ratcliff R
    5. Brown SD

    (2015) Revisiting the evidence for collapsing boundaries and urgency signals in perceptual decision-making

    Journal of Neuroscience 35:2476–2484.

    https://doi.org/10.1523/JNEUROSCI.2410-14.2015

    • PubMed
    • Google Scholar
19. 1. Hosokawa T
    2. Kennerley SW
    3. Sloan J
    4. Wallis JD

    (2013) Single-neuron mechanisms underlying cost-benefit analysis in frontal cortex

    Journal of Neuroscience 33:17385–17397.

    https://doi.org/10.1523/JNEUROSCI.2221-13.2013

    • PubMed
    • Google Scholar
20. 1. Huang HJ
    2. Kram R
    3. Ahmed AA

    (2012) Reduction of metabolic cost during motor learning of arm reaching dynamics

    Journal of Neuroscience 32:2182–2190.

    https://doi.org/10.1523/JNEUROSCI.4003-11.2012

    • PubMed
    • Google Scholar
21. 1. Izawa J
    2. Rane T
    3. Donchin O
    4. Shadmehr R

    (2008) Motor adaptation as a process of reoptimization

    Journal of Neuroscience 28:2883–2891.

    https://doi.org/10.1523/JNEUROSCI.5359-07.2008

    • PubMed
    • Google Scholar
22. Book

    1. Macmillan NA
    2. Creelman CD

    (2005)

    Detection theory

    Mahwah, New Jersey, London: A User’s Guide Lawrence Erlbaum associates.

    • Google Scholar
23. 1. Makin TR
    2. Wilf M
    3. Schwartz I
    4. Zohary E

    (2010) Amputees "neglect" the space near their missing hand

    Psychological Science 21:55–57.

    https://doi.org/10.1177/0956797609354739

    • PubMed
    • Google Scholar
24. 1. Marcos E
    2. Cos I
    3. Girard B
    4. Verschure PF

    (2015) Motor cost influences perceptual decisions

    PLoS One 10:e0144841.

    https://doi.org/10.1371/journal.pone.0144841

    • PubMed
    • Google Scholar
25. 1. Mather G
    2. Pavan A
    3. Campana G
    4. Casco C

    (2008) The motion aftereffect reloaded

    Trends in Cognitive Sciences 12:481–487.

    https://doi.org/10.1016/j.tics.2008.09.002

    • PubMed
    • Google Scholar
26. 1. Mazzoni P
    2. Hristova A
    3. Krakauer JW

    (2007) Why don't we move faster? Parkinson's disease, movement vigor, and implicit motivation

    Journal of Neuroscience 27:7105–7116.

    https://doi.org/10.1523/JNEUROSCI.0264-07.2007

    • PubMed
    • Google Scholar
27. 1. Mulder MJ
    2. Wagenmakers EJ
    3. Ratcliff R
    4. Boekel W
    5. Forstmann BU

    (2012) Bias in the brain: a diffusion model analysis of prior probability and potential payoff

    Journal of Neuroscience 32:2335–2343.

    https://doi.org/10.1523/JNEUROSCI.4156-11.2012

    • PubMed
    • Google Scholar
28. 1. Oliveira FT
    2. Diedrichsen J
    3. Verstynen T
    4. Duque J
    5. Ivry RB

    (2010) Transcranial magnetic stimulation of posterior parietal cortex affects decisions of hand choice

    PNAS 107:17751–17756.

    https://doi.org/10.1073/pnas.1006223107

    • PubMed
    • Google Scholar
29. 1. Palmer J
    2. Huk AC
    3. Shadlen MN

    (2005) The effect of stimulus strength on the speed and accuracy of a perceptual decision

    Journal of Vision 5:1–404.

    https://doi.org/10.1167/5.5.1

    • PubMed
    • Google Scholar
30. 1. Prévost C
    2. Pessiglione M
    3. Météreau E
    4. Cléry-Melin ML
    5. Dreher JC

    (2010) Separate valuation subsystems for delay and effort decision costs

    Journal of Neuroscience 30:14080–14090.

    https://doi.org/10.1523/JNEUROSCI.2752-10.2010

    • PubMed
    • Google Scholar
31. 1. Rahnev D
    2. Maniscalco B
    3. Graves T
    4. Huang E
    5. de Lange FP
    6. Lau H

    (2011) Attention induces conservative subjective biases in visual perception

    Nature Neuroscience 14:1513–1515.

    https://doi.org/10.1038/nn.2948

    • PubMed
    • Google Scholar
32. 1. Ratcliff R
    2. McKoon G

    (2008) The diffusion decision model: theory and data for two-choice decision tasks

    Neural Computation 20:873–922.

    https://doi.org/10.1162/neco.2008.12-06-420

    • PubMed
    • Google Scholar
33. 1. Reichenbach A
    2. Costello A
    3. Zatka-Haas P
    4. Diedrichsen J

    (2013) Mechanisms of responsibility assignment during redundant reaching movements

    Journal of Neurophysiology 109:2021–2028.

    https://doi.org/10.1152/jn.01052.2012

    • PubMed
    • Google Scholar
34. 1. Resulaj A
    2. Kiani R
    3. Wolpert DM
    4. Shadlen MN

    (2009) Changes of mind in decision-making

    Nature 461:263–266.

    https://doi.org/10.1038/nature08275

    • PubMed
    • Google Scholar
35. 1. Schwarz G

    (1978) Estimating the dimension of a model

    The Annals of Statistics 6:461–464.

    https://doi.org/10.1214/aos/1176344136

    • Google Scholar
36. 1. Shadlen MN
    2. Hanks TD
    3. Churchland AK
    4. Kiani R
    5. Yang T

    (2006)

    Bayesian Brain: Probabilistic Approaches to Neural Coding

    The speed and accuracy of a simple perceptual decision: a mathematical primer, Bayesian Brain: Probabilistic Approaches to Neural Coding, Cambridge, MIT Press.

    • Google Scholar
37. 1. Shadlen MN
    2. Kiani R
    3. Hanks TD
    4. Churchland AK

    (2008)

    Better Than Conscious?: Decision Making, the Human Mind, and Implications For Institutions

    Neurobiology of decision making: an intentional framework, Better Than Conscious?: Decision Making, the Human Mind, and Implications For Institutions, Cambridge, MIT Press.

    • Google Scholar
38. 1. Simen P
    2. Contreras D
    3. Buck C
    4. Hu P
    5. Holmes P
    6. Cohen JD

    (2009) Reward rate optimization in two-alternative decision making: Empirical tests of theoretical predictions

    Journal of Experimental Psychology: Human Perception and Performance 35:1865–1897.

    https://doi.org/10.1037/a0016926

    • Google Scholar
39. 1. Smith MA
    2. Ghazizadeh A
    3. Shadmehr R

    (2006) Interacting adaptive processes with different timescales underlie short-term motor learning

    PLoS Biology 4:e179.

    https://doi.org/10.1371/journal.pbio.0040179

    • PubMed
    • Google Scholar
40. 1. Sun P
    2. Landy MS

    (2016) A Two-Stage process model of sensory discrimination: an alternative to Drift-Diffusion

    Journal of Neuroscience 36:11259–11274.

    https://doi.org/10.1523/JNEUROSCI.1367-16.2016

    • PubMed
    • Google Scholar
41. 1. Tajima S
    2. Drugowitsch J
    3. Pouget A

    (2016) Optimal policy for value-based decision-making

    Nature Communications 7:12400.

    https://doi.org/10.1038/ncomms12400

    • PubMed
    • Google Scholar
42. 1. Thoroughman KA
    2. Shadmehr R

    (2000) Learning of action through adaptive combination of motor primitives

    Nature 407:742–747.

    https://doi.org/10.1038/35037588

    • PubMed
    • Google Scholar
43. 1. Wagenmakers EJ
    2. Farrell S

    (2004) AIC model selection using akaike weights

    Psychonomic Bulletin & Review 11:192–196.

    https://doi.org/10.3758/BF03206482

    • PubMed
    • Google Scholar
44. 1. Webster MA

    (2011) Adaptation and visual coding

    Journal of Vision 11:3.

    https://doi.org/10.1167/11.5.3

    • PubMed
    • Google Scholar
45. 1. Whiteley L
    2. Sahani M

    (2008) Implicit knowledge of visual uncertainty guides decisions with asymmetric outcomes

    Journal of Vision 8:2–15.

    https://doi.org/10.1167/8.3.2

    • PubMed
    • Google Scholar

Author details

1. Nobuhiro Hagura

   1. Institute of Cognitive Neuroscience, University College London, London, United Kingdom
   2. Center for Information and Neural Networks (CiNet), National Institute of Communications and Technology, Suita City, Japan

   Contribution

   NH, Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Writing—original draft, Writing—review and editing

   For correspondence

   n.hagura@nict.go.jp

   Competing interests

   The authors declare that no competing interests exist.

   "This ORCID iD identifies the author of this article:" 0000-0001-9852-5056

2. Patrick Haggard

   Institute of Cognitive Neuroscience, University College London, London, United Kingdom

   Contribution

   PH, Conceptualization, Supervision, Funding acquisition, Writing—review and editing

   Competing interests

   The authors declare that no competing interests exist.

3. Jörn Diedrichsen

   1. Institute of Cognitive Neuroscience, University College London, London, United Kingdom
   2. Brain and Mind Institute, Western University, London, Canada

   Contribution

   JD, Conceptualization, Resources, Supervision, Funding acquisition, Methodology, Writing—review and editing

   Competing interests

   The authors declare that no competing interests exist.

   "This ORCID iD identifies the author of this article:" 0000-0003-0264-8532

• 11,691

  views

• 1,416

  downloads

• 72

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.