Trade-Off Between Memory Usage and Optimization Speed in a Simple Machine Learning Task

In two experiments the effect of loud noise on a simple vibrotactile learning task was studied. After learning the task to criterion, 10 male and 10 female Ss received two pairs of test trials, one without noise, and one in continuous noise (a 1,000 cps pure tone at 90 dB SPL). An additional 10 male and 10 female Ss learned the same task and also received the same two pairs of test trials, but instead of receiving continuous noise for the second condition, they received an intermittent noise (random numbers presented at 2 sec intervals at 95 dB SPL). In the first experiment noise had a significant effect on the performance of the 20-subject group and also on the females in the group. Noise did not significantly affect the performance of males. In the second experiment noise had no significant effect on either males or females.

This study tested the extent to which the presence and relevance of interactive features influence toddlers’ visual attention and learning from videos. Two-year-olds (N = 88) watched an animated bear hide on video and then searched for a bear sticker on a corresponding felt board. To watch the bear hide, toddlers watched without touching the screen (non-interactive condition), touched a relevant part of the screen (the bear; relevant condition), or touched an irrelevant part of the screen (a star; irrelevant condition). A subset of 60 children (68%) wore a head-mounted eye-tracker to record their eye movements while watching the bear hide. A relevant touchscreen feature did not increase toddlers’ learning compared to non-interactive video, despite increasing toddlers’ visual attention to target information. Conversely, the irrelevant touchscreen feature had a positive effect on learning, particularly on later search trials. The findings are discussed with respect to toddlers’ mental representations of on-screen versus real-life events and implications for the design of touchscreen media for children. IMPACT SUMMARY a. Prior State of Knowledge: Toddlers learn less from video than from real-life experiences. It is widely believed that interactive media support toddlers’ learning, but research is mixed, sometimes finding neutral or even negative effects of interactivity compared to noninteractive video. b. Novel Contributions: A simple, relevant touchscreen feature increased toddlers’ encoding of target information but did not improve learning. Conversely, an irrelevant touchscreen feature improved toddlers’ learning, perhaps by increasing slightly the effort required to complete an otherwise simple task. c. Practical Implications: Interactive media do not always increase toddlers’ learning. Simple learning tasks may not benefit from simple interactivity. Instead, media creators may be able to maximize educational value by balancing the difficulty of the lesson with the complexity of media features.

Drawing by learners can be an effective way to develop memory and generate visual models for higher-order skills in biology, but students are often reluctant to adopt drawing as a study method. We designed a nonclassroom intervention that instructed introductory biology college students in a drawing method, minute sketches in folded lists (MSFL), and allowed them to self-assess their recall and problem solving, first in a simple recall task involving non-European alphabets and later using unfamiliar biology content. In two preliminary ex situ experiments, students had greater recall on the simple learning task, non-European alphabets with associated phonetic sounds, using MSFL in comparison with a preferred method, visual review (VR). In the intervention, students studying using MSFL and VR had ∼50-80% greater recall of content studied with MSFL and, in a subset of trials, better performance on problem-solving tasks on biology content. Eight months after beginning the intervention, participants had shifted self-reported use of drawing from 2% to 20% of study time. For a small subset of participants, MSFL had become a preferred study method, and 70% of participants reported continued use of MSFL. This brief, low-cost intervention resulted in enduring changes in study behavior.

Individual differences in learning ability are negatively linked to behavioural plasticity in a frequency-dependent game

The IntelliCage (IC) permits the assessment of the behavior and learning abilities of mice in a social home cage context. To overcome water deprivation as an aversive driver of learning, we developed protocols in which spatial learning is motivated appetitively by the preference of mice for sweetened over plain water. While plain water is available at all times, only correct task responses give access to sweetened water rewards. Under these conditions, C57BL/6J mice successfully mastered a corner preference task with the reversal and also learned a more difficult time-place task with reversal. However, the rate of responding to sweetened water decreased strongly with increasing task difficulty, indicating that learning challenges and reduced success in obtaining rewards decreased the motivation of the animals to seek sweetened water. While C57BL/6J mice of both sexes showed similar initial taste preferences and learned similarly well in simple learning tasks, the rate of responding to sweetened water and performance dropped more rapidly in male than in female mice in response to increasing learning challenges. Taken together, our data indicate that male mice can have a disadvantage relative to females in mastering difficult, appetitively motivated learning tasks, likely due to sex differences in value-based decision-making.

Cataracts are a common visual impairment that directly affects visual acuity and contrast sensitivity, both of which constrain mid-level and high-level visual processes. Fortunately, cataracts are also relatively easy to treat, so many adults with impaired vision caused by cataract can have their vision restored successfully. This raises the question of how impaired vision due to cataract onset and the restoration of normal vision via surgery impact performance on recognition tasks. In the present study, we examined this question in the context of learning to recognize new faces: How well do observers establish useful representation of novel face identities in the context of cataracts and how do changes in vision associated with cataract onset and treatment affect recognition of recently-learned faces? We used cataract simulation goggles to vary the appearance of face stimuli during the learning and test phases of a simple novel face learning task and examined how recognition for learned and unlearned faces was affected by visual quality in each learning phase.

Associative Learning![Figure][1] Nematode worm, Caenorhabditis elegans PHOTO: SINCLAIR STAMMERS/SCIENCE SOURCE In associative learning, you link potentially unrelated things because you are exposed to them at the same time. Ohno et al. studied a simple associative learning task in the nematode worm Caenorhabditis elegans . They presented the worms with a taste substance while withholding food. After starving in the presence of the taste substance, the animals switched their behavior from being attracted to the taste to finding it aversive. A specific isoform of the insulin receptor is critical for this type of associative learning—at least in worms. Science , this issue p. [313][2] [1]: pending:yes [2]: /lookup/volpage/345/313?iss=6194

Many decades after the introduction of ICT into classrooms there are still unanswered questions about the impact of technology in the long and short term on students' learning, and how it has affected simple and complex learning tasks. These are important for (a) forming government policies; (b) directing teacher education programmes: (c) advancing national curricula; (d); designing or reforming classroom implementation and (e) analysing costs and benefits. While a plethora of studies has been conducted on the effects of ICT in education, major policy and methodological problems have precluded an unambiguous answer to such questions as:--"Does the way in which ICT is implemented have a major/minor impact on students' knowledge and understanding?" and "Does the impact affect the surface or deep structure of students' thinking and acting?" To date we have had no large-scale longitudinal studies of ICT's impact in education such as we have in the form of studies of earlier major curriculum development projects. Nor have we had many comprehensive studies of the complex interactions between various types of ICT implementation and the effects of other factors such as school-based interventions, socio-economic status and school expenditures which have been shown to have a greater impact on education compared with other previous innovations in education. Furthermore we do not know if previous research studies have used research methods that matched learning objectives to instruments/procedures. Many previous studies are vague as to the actual measures used but we can infer that standardized tests were a frequent measure. In other instances, ad hoc analyses, with criteria that may have varied from analyst to analyst and were not "blind" analyses were certainly used to measure "success." All of these limitations and uncertainties and many more point to the need for a thorough, rigorous, and multifaceted approach to analysing the impact of ICT on students' learning. This paper draws on previous research evidence to identify relevant research strategies to address the gaps in our knowledge about ICT and students' learning explained above.

It has been suggested that people and nonhuman animals protect their knowledge from interference by shifting attention toward the context when presented with information that contradicts their previous beliefs. Despite that suggestion, no studies have directly measured changes in attention while participants are exposed to an interference treatment. In the present experiments, we adapted a dot-probe task to track participants’ attention to cues and contexts while they were completing a simple category learning task. The results support the hypothesis that interference produces a change in the allocation of attention to cues and contexts.

P.6.a.001 The difference of decision-making pattern under ambiguity and risk between alcohol dependence and schizophrenia

It has long been known that in some relatively simple reinforcement learning tasks traditional strength-based classifier systems will adapt poorly and show poor generalisation. In contrast, the more recent accuracy-based XCS, appears both to adapt and generalise well. In this work, we attribute the difference to what we call strong over general and fit over general rules. We begin by developing a taxonomy of rule types and considering the conditions under which they may occur. In order to do so an extreme simplification of the classifier system is made, which forces us toward qualitative rather than quantitative analysis. We begin with the basics, considering definitions for correct and incorrect actions, and then correct, incorrect, and overgeneral rules for both strength and accuracy-based fitness. The concept of strong overgeneral rules, which we claim are the Achilles' heel of strength-based classifier systems, are then analysed. It is shown that strong overgenerals depend on what we call biases in the reward function (or, in sequential tasks, the value function). We distinguish between strong and fit overgeneral rules, and show that although strong overgenerals are fit in a strength-based system called SB-XCS, they are not in XCS. Next we show how to design fit overgeneral rules for XCS (but not SB-XCS), by introducing biases in the variance of the reward function, and thus that each system has its own weakness. Finally, we give some consideration to the prevalence of reward and variance function bias, and note that non-trivial sequential tasks have highly biased value functions.

The amount of feedback needed to learn a new task has long been debated ( cf. Schmidt & Bjork, 1992; Van Merrienboer & Sweller, 2005). One potential answer to the amount of feedback required is that feedback should be contingent upon the learner's cognitive resources and demands imposed by the task (McLaughlin, Rogers, & Fisk, 2006). To test this model, a study is proposed that accounts for the learner's cognitive resources by comparing samples of populations with known differences, older and younger adults (Horn & Cattell, 1967; Salthouse & Babcock, 1991). To account for task demands, a simple rule-based cue learning task has been created. Participants will be provided with varying levels of feedback specificty while learning the task. We predict younger adults will benefit from less feedback support while older adults benefit from more feedback support. Theoretical and applied contributions are also discussed.

Some neurons have stimulus responses that are stable over days, whereas other neurons have highly plastic stimulus responses. Using a recurrent network model, we explore whether this could be due to an underlying diversity in their synaptic plasticity. We find that, in a network with diverse learning rates, neurons with fast rates are more coupled to population activity than neurons with slow rates. This plasticity-coupling link predicts that neurons with high population coupling exhibit more long-term stimulus response variability than neurons with low population coupling. We substantiate this prediction using recordings from the Allen Brain Observatory, finding that a neuron's population coupling is correlated with the plasticity of its orientation preference. Simulations of a simple perceptual learning task suggest a particular functional architecture: a stable 'backbone' of stimulus representation formed by neurons with low population coupling, on top of which lies a flexible substrate of neurons with high population coupling.

Three distinct nuclei of the chick forebrain--the intermediate medial hyperstriatum ventrale (IMHV), lobus parolfactorius (LPO), and paleostriatum augmentatum (PA)--show metabolic, morphological, and neurophysiological changes following training on a passive avoidance task, suggesting that these and other areas of the chick forebrain participate in memory formation for this task. Considerable evidence exists for lateralization of memory processes in the chick. Several experiments examined the effects of lesions in the IMHV on the ability of chicks to learn and retain the avoidance task. Pre-training bilateral lesions in the IMHV produced an impairment in avoidance responding tested three hours after training. Pre-training unilateral lesions in the left but not the right IMHV resulted in a similar impairment. However, bilateral IMHV ablations, given either 1 or 6 h post-training, did not impair retention. IMHV lesions did not impair retention of a simple escape learning task. These results are consistent with other studies that have examined the effects of bilateral IMHV lesions on acquisition of passive avoidance and extend these findings by demonstrating lateralization of acquisition involving the left IMHV. The results also suggest that, as early as one hour post-training, the IMHV is not necessary to retain the memory and indicate that other forebrain structures, possibly the LPO or PA, may maintain the memory trace following training. Hypotheses to account for these results and indications of future research are discussed.

Evolutionary learning in reaction-diffusion neurons