Attention, Executive Control and Decisions

A monkey watches a cloud of dots, some moving together and the rest at random, and must decide which way the coherent dots are going. As it watches, a neuron in its parietal cortex fires faster and faster, and when the firing reaches a certain level the monkey moves its eyes to give its answer. Is that neuron making the decision? Silence the area it sits in and the monkey decides just as well. This lesson builds toward that puzzle through three abilities that sit between seeing and acting: holding information in mind, choosing what to process, and turning evidence into a choice.

Holding and using information: prefrontal cortex keeps the goal in view

Executive control is the ability to organize thought and action according to internal goals rather than whatever the senses offer next. Working memory, holding information that is no longer in front of you so that you can use it, is one of its tools. Both are tied to the prefrontal cortex, the front part of the frontal lobe, which never works alone. Working memory is different from the long-term stores in memory systems.

The classic recording came from monkeys. Funahashi, Bruce and Goldman-Rakic (1989) trained them in an oculomotor delayed-response task. The monkey fixated a central spot, a cue flashed for 0.5 s at one of several peripheral places, and the monkey kept fixating through a delay of 1 to 6 s before making a saccade (a fast eye movement) to where the cue had been. Fixation is the key control: with the eyes still and the screen blank, firing that depends on the cue’s location cannot be a response to a visible cue or an eye movement under way (a planned movement is not ruled out by fixation alone).

Of 288 neurons recorded around the principal sulcus, 87 changed their firing significantly during the delay, and 79% of those 87 were directionally selective. Do the arithmetic before you summarize it: 0.79 × 87 ≈ 69 cells, and 69 / 288 ≈ 24% of the sample. A subset of principal-sulcus neurons carried the remembered direction, in this task, in monkeys.

Prefrontal neurons also code what to do. Wallis, Anderson and Miller (2001) trained monkeys to judge two pictures under one of two rules, respond if they match or respond if they differ. The most common activity in their prefrontal sample reflected the rule in force, and the monkeys applied the rules to pictures they had never seen. The new pictures matter: a cell that signals the rule for a novel picture cannot be signalling a memorized picture-response pairing.

Miller and Cohen (2001) proposed a theory that ties these findings together: prefrontal cortex maintains representations of goals, and these bias processing in other structures toward what the goal needs. Think of a referee’s whistle held up front, tilting contests that happen elsewhere. It is a proposal, and the prefrontal cortex is one partner among several.

Two newer results widen the picture. In mice running a delayed non-match-to-sample task in a T-maze, Bolkan and colleagues (2017) inhibited pathways selectively. Input from the mediodorsal thalamus (a thalamic nucleus, or cluster of neurons, connected with prefrontal cortex) to medial prefrontal cortex supported maintenance during the delay, and the reverse projection supported the later choice. Delay firing in medial prefrontal cortex needed the thalamic input, yet those delay-tuned neurons overlapped minimally with spatially tuned neurons, so mouse delay activity did not simply copy the monkey result. In humans, Rose and colleagues (2016) cued people to set one remembered item aside. That item’s category could no longer be decoded from the EEG, but a single TMS pulse briefly made it decodable again, and only while it might still be needed. The authors suggest that short-term synaptic changes could hold such latent items; the synapses were not measured, and because decoding was at the category level, reinstated category context is an alternative to reactivation of the item itself.

Attention changes what neurons report

Attention is selecting part of what is available for further processing. Keep it apart from arousal, the overall level of alertness, which shifts with sleep and rhythms. A receptive field is the region of space in which a stimulus changes a neuron’s firing, as in visual cortex.

Moran and Desimone (1985) recorded from neurons in macaque visual area V4 and placed two stimuli inside one receptive field: one that drove the cell well and one that drove it poorly. When the monkey attended to the poor stimulus, the response fell toward what that stimulus gave alone, although the retinal image was unchanged. In primary visual cortex, cells were unaffected by attention. Stimuli seem to compete for a neuron, and attention biases the competition.

Attention is not one fixed multiplier. Reynolds, Pasternak and Desimone (2000) varied contrast and found that attention raised V4 responses more at low contrast than at high. On average it acted like multiplying contrast by 1.51: an attended 10% contrast grating drove cells about as well as an unattended grating of 10 × 1.51 ≈ 15%. Near the top of the contrast-response curve, where responses saturate, extra effective contrast changes little.

Attention also changes how neurons vary together. Noise correlation is the correlation of two neurons’ spike counts across repeats of the same stimulus:

ρ = Cov(n1, n2) / √(Var(n1) × Var(n2))

Kanashiro and colleagues (2017) reanalyzed V4 recordings from two monkeys detecting orientation changes. Mean noise correlation was 0.065 unattended and 0.045 attended. Mean covariance fell 34.0% while mean single-neuron variance rose 1.61%. Private noise partly cancels when a downstream cell pools many inputs; shared noise does not, so less of it makes the pool more reliable. Their circuit model predicts that attention acts mainly on inhibitory neurons, a prediction still to be tested.

Damage shows the same networks at work. After damage to parietal or frontal cortex, a patient may fail to notice or respond to things on the side of space opposite the lesion (contralesional hemispatial neglect). Neglect is commoner and more severe after right-hemisphere damage.

Accumulating evidence toward a choice: a ramp to a bound

In the random-dot task, coherence is the percentage of dots moving together; the rest move randomly. Neurons in area MT are tuned to direction (see sensory coding). Newsome, Britten and Movshon (1989) found that most of the MT neurons they studied, on a task matched to the neurons’ properties, discriminated motion about as well as, or better than, the monkey. Salzman, Britten and Newsome (1990) then stimulated small clusters of MT neurons with a shared preferred direction and biased the monkeys’ choices toward that direction. MT supplies evidence, and its activity can sway the choice.

The lateral intraparietal area (LIP) in the parietal lobe holds neurons that fire before saccades into their response field. Roitman and Shadlen (2002) let rhesus monkeys answer whenever they were ready, with one choice target in the recorded neuron’s response field. Stronger motion gave faster and more accurate choices. LIP firing ramped, predicted the choice, rose more steeply with stronger motion, and reached a similar level shortly before the saccade, which the authors read as a threshold that appeared to mark the end of the decision. These ramps are averages across trials. Whether single trials ramp or step is debated.

A drift-to-bound model captures this: momentary evidence is added up until the total hits a bound, and the drift, the average slope, grows with coherence.

rise time = (bound − baseline) / slope reaction time = latency + rise time + motor delay

Worked problem. The numbers are invented for teaching. An LIP neuron fires 30 spikes/s until 200 ms after motion onset, then rises linearly at 7.8125 (spikes/s) per second for each percent coherence. The decision occurs at 70 spikes/s, and the saccade follows 70 ms later. Assume no noise and a single neuron standing for the accumulator.

Step 1, slopes. 7.8125 × 51.2 = 400 (spikes/s)/s; × 12.8 = 100; × 3.2 = 25.

Step 2, rise times. The distance to the bound is 70 − 30 = 40 spikes/s. 40 / 400 = 0.10 s; 40 / 100 = 0.40 s; 40 / 25 = 1.60 s. Units: (spikes/s) / ((spikes/s)/s) = s.

Step 3, reaction times. 200 + 100 + 70 = 370 ms at 51.2%; 200 + 400 + 70 = 670 ms at 12.8%; 200 + 1,600 + 70 = 1,870 ms at 3.2%.

Step 4, compare. In one illustrated experiment, Roitman and Shadlen reported a mean of 350 ± 9 ms (SEM) for correct choices at 51.2%, close to the model. At 0% the mean slope is zero, so only noise reaches the bound: long, variable times and chance choices. In a noisy model with one bound for each choice, noise carries many weak-motion trials to one bound or the other early, so it predicts shorter mean times at weak motion than this noiseless version.

trials reaching the preferred-target bound trials reaching the other-choice bound noiseless prediction observed mean expected mean, this noisy sketch

Motion coherence 12.8% Noise 0 (relative units; 0 = none) Resample (200 trials)

The lesson’s arithmetic

—

Noiseless reaction time (expected)

—

Observed mean, noisy trials

—

Expected mean for this noisy sketch

—

Choices toward the preferred target

—

—

In this sketch: invented numbers (30 spikes/s baseline, 70 spikes/s bound, 200 ms latency, 70 ms motor delay), one neuron standing for the accumulator, no noise unless you set it. With noise, the accumulator takes random steps in relative units and a second bound for the other choice sits 40 relative units below the baseline, symmetric with the upper bound, which is why it can fall below 0 on this relative scale. Time is cut off at 3 s on the plot; noisy trials are run to 6 s. Expected means and choice percentages are the standard first-passage results for this random walk between two bounds, not figures from the lesson.

rise time = (bound − baseline) / slope, and reaction time = latency + rise time + motor delay. The ramps in the lesson are trial averages, and whether single trials ramp or step is debated; this sketch draws a noisy accumulator and does not settle that.

How long to keep sampling also depends on what a choice is worth, which links to motivation and reward, and selecting one action among many draws on the basal ganglia.

When the best correlate is not necessary

Correlation shows what activity goes with a choice; a manipulation shows whether the choice needs it. Katz, Yates, Pillow and Huk (2016) recorded MT and LIP in rhesus monkeys and then reversibly silenced each with muscimol, a GABA-A receptor agonist. MT inactivation impaired motion discrimination. LIP inactivation produced no measurable impairment of motion decisions in the tasks tested. In a free-choice task, the same LIP inactivation changed spatial selection and eye-movement metrics.

Design problem

Muscimol in MT raises the motion-discrimination threshold; in LIP it leaves the threshold unchanged, though monkeys choose targets in the inactivated field less often in a free-choice task. What does this say about LIP as “the accumulator”, and what would you do next?

One way to do it

LIP activity correlates with the decision, but LIP was not necessary for the discrimination under these conditions. The free-choice effect shows the inactivation worked, so the null result is not a missed injection. MT was needed for this discrimination, as expected of a stage that carries direction evidence, though direction-selective signals already reach MT from V1.

Several readings remain. Other areas that carry choice signals, such as the frontal eye field or the superior colliculus, may hold the signal redundantly or compensate when LIP is silenced. Or LIP may reflect a decision computed elsewhere. Next, inactivate several areas together, and record many neurons at once to see single-trial dynamics. Scope every conclusion to this task, this training and this species. The circuit experiment design and large-scale neuroscience lessons build those tools.