About Brainglot
Brainglot personnel
Research Areas and Projects
Consolider CogNeuro Seminar Series
Consolider Meetings and Workshops
Technical and Administrative Support Services
Research Reports
Consolider Groups
Speech Perception, Production and Bilingualism
Cognitive Neuroscience of Auditory Perception and Attention
Group of Attention, Action and Perception
Computational and Theoretical Neuroscience
Neuropsychology and Functional Neuroimaging
Grammar and Bilingualism

Multiple-choice decision-making

Project coordinator: Gustavo Deco.

Project members: Larissa Albantakis.                                                                                                          


Brain processes during decision-making are a topic of unceasing interest. Building on a large foundation of experimental and theoretical studies on decision-making in its simplest form (i.e. binary perceptual choices) this project investigates via a theoretical/computational approach how the mechanisms underlying decision-formation adapt to choices between more than two alternatives. In particular, we make use of biophysically realistic computational models of brain regions involved in the decision-making process to address the following questions: Can the same (neural) mechanisms account for binary and multiple-choice decision-making? How are the different choice alternatives encoded in the brain? Which parameters are affected by the number of alternatives? Is there a need for regulatory brain signals that explicitly encode the number of alternatives, or are the brain regions involved in decision-making able implicitly to adjust to the number of choice targets? How do experimentally observed relationships between reaction time and the number of choice alternatives emerge based on neural processes?


SUBPROJECT: A computational framework of multiple-choice decision-makingThis subproject addresses the question of how different numbers of choice alternatives might be handled and encoded in the brain. Several behavioural studies on multiple-choice decision-making have shown a logarithmic relationship between the number of alternatives and reaction time, which became known as “Hick’s Law”. However, the neurological origin of this relationship remains unknown. Neurophysiological data recorded during a perceptual multiple-choice decision task was first presented only recently when macaque LIP neurons were recorded during a random-dot motion task with two and four alternatives (see Figure 1 for a sketch of the experimental paradigm). Based on this experimental evidence we aim to build a biophysically realistic computational model for multiple-choice processes in order to provide mechanistic explanations of the observed behaviour and neural activity.

Extending computational models of binary choices to multiple alternatives. We have so far extended a well-established biophysically detailed attractor model on binary decision-making to up to four possible alternatives. The network is implemented with integrate-and-fire neurons, which have three types of receptors (AMPA, NMDA and GABA). As the receptor dynamics are explicitly modelled using physiological parameters, the model is capable of reproducing, explaining and predicting neural firing rates, in our case, of LIP neurons. Each of the four choice alternatives is encoded by a population of excitatory neurons that responds selectively to motion towards one of the choice targets. A population of inhibitory neurons provides global inhibitory feedback and enables the network to perform winner-take-all decision-making. The extended model accounts for all relevant aspects of the above-mentioned neuro physiological study on both the cellular and behavioural levels. It reproduces the main experimental finding that the same decision-threshold applies independently of the number of alternatives, although the starting point of the ramping activity is lower for four than for two choice alternatives. Moreover, the model explains longer reaction times in the case of two targets with 90° instead of 180° angular separation, as the firing rate transients in the former take longer to split than in the 180° case (Figure 1).

Encoding of choice alternatives. Notably, the observed differences in firing rate between the tested experimental paradigms (two and four alternatives and two alternatives with an angular separation of 90°) emerge implicitly in the network due to feedback inhibition without extra top-down regulation mechanisms to adapt the network to the number of choices. The network thus exhibits categorical decision-making for two and four choices for the same range of external inputs. Previously, networks with discrete populations have been adjusted to exhibit winner-take-all competition for one particular set of choice alternatives or memory states. In our study, we analysed how the network’s competition regimes could be brought into accord for different numbers of alternatives. In a mean-field approximation of the network we found a linearly increasing relationship between the relative size of the selective neural populations and the common range of decision-making for two and four alternatives. This implies that pooling over many neurons favours choice-number independent decision-making. Our results consequently suggest a physiological advantage of a pooled, multineuron representation of choice alternatives. Extending the current model to even more alternatives is a future objective, particularly with regard to Hick’s Law.





Figure 1. Experimental paradigm of the random-dot motion task and single trial examples of simulated LIP activity during the choice process for each of the three experimental conditions (two and four alternatives and two alternatives with an angular separation of 90°). Each choice alternative is encoded in a population of neurons selective for motion towards the respective choice target (mean firing rates are coloured according to left sketch). Red: inhibitory population of neurons, black: nonselective neurons.

  1. Analytical approach to continuous decision-making, based on neural field models to investigate the neurodynamical mechanisms underlying the Hick's law
  2. Changes of mind in a multiple-choice attractor model of decision making (see Project 9)